'); 48 | }; 49 | 50 | Utils.getUrlFromDragEvent = function getUrlFromDragEvent(e) { 51 | var elem = e.originalEvent.dataTransfer.getData('text/html'); 52 | var url = $(elem).find('img').addBack('img').attr('src'); 53 | if (!url) { 54 | url = $(elem).find('a').addBack('a').attr('href'); 55 | } 56 | if (!url) { 57 | url = e.originalEvent.dataTransfer.getData('text/plain'); 58 | } 59 | return url; 60 | }; 61 | 62 | /** 63 | * Compile the Handlebars templates 64 | */ 65 | Utils.compileTemplates = function compileTemplates() { 66 | var templates = {}; 67 | $("*[id^='tpl-']").each(function() { 68 | var $this = $(this); 69 | var tplId = $this.attr('id').replace(/^tpl-/, ''); 70 | templates[tplId] = Handlebars.compile($this.html()); 71 | }); 72 | return templates; 73 | }; 74 | 75 | /** 76 | * Shrink/expand a textarea to fit its contents 77 | */ 78 | Utils.fitHeight = function fitHeight(selector) { 79 | $(selector).each(function() { 80 | $(this) 81 | .attr('rows', 1) // Must be one for single-line textareas 82 | .css({'height': 'auto', 'overflow-y': 'hidden', 'resize': 'none'}) 83 | .height(this.scrollHeight); 84 | }); 85 | }; 86 | -------------------------------------------------------------------------------- /js/views/selectbar.js: -------------------------------------------------------------------------------- 1 | function Selectbar(opts) { 2 | for (let key in opts) { this[key] = opts[key]; } 3 | this.$el = $(this.el); 4 | } 5 | 6 | Selectbar.prototype.enter = function enter() { 7 | this.selectLines('unselect'); 8 | window.app.selectMode = true; 9 | this.$el.removeClass('hidden'); 10 | window.app.emit('app:enter-select-mode'); 11 | }; 12 | 13 | Selectbar.prototype.exit = function exit() { 14 | this.selectLines('unselect'); 15 | window.app.selectMode = false; 16 | this.$el.addClass('hidden'); 17 | window.app.emit('app:exit-select-mode'); 18 | }; 19 | 20 | Selectbar.prototype.toggle = function toggle() { 21 | this[window.app.selectMode ? 'exit' : 'enter'](); 22 | }; 23 | 24 | Selectbar.prototype.getSelection = function getSelection() { 25 | var ret = []; 26 | for (let i = 0; i < app.pageView.lineViews.length; i++) { 27 | var lineView = app.pageView.lineViews[i]; 28 | if (lineView.selected) ret.push(lineView); 29 | } 30 | return ret; 31 | }; 32 | 33 | Selectbar.prototype.selectLines = function selectLines(action, ids) { 34 | var app = window.app; 35 | // If no id was passed, use all ids 36 | if (!ids) { 37 | ids = []; 38 | for (let i = 0; i < app.currentPage.lines.length; i++) { 39 | ids.push(i); 40 | } 41 | } 42 | for (let i = 0; i < ids.length; i++) { 43 | var lineView = app.pageView.lineViews[ids[i]]; 44 | lineView.selected = (action === 'select' ? true : action === 'unselect' ? false : !lineView.selected); 45 | lineView.renderCheckbox(); 46 | } 47 | }; 48 | 49 | Selectbar.prototype.render = function renderSelectBar() { 50 | var self = this; 51 | var app = window.app; 52 | 53 | this.$el.find('.select-all').on('click', function selectAll() { 54 | self.selectLines('select'); 55 | }); 56 | this.$el.find('.select-none').on('click', function selectNone() { 57 | self.selectLines('unselect'); 58 | }); 59 | this.$el.find('.select-toggle').on('click', function selectToggle() { 60 | self.selectLines('toggle'); 61 | }); 62 | this.$el.find('*[data-tag]').on('click', function addTagMultiple() { 63 | var tag = $(this).attr('data-tag'); 64 | var selection = self.getSelection(); 65 | for (let i = 0; i < selection.length; i++) { 66 | selection[i].addTag(tag); 67 | } 68 | window.app.emit('app:changed'); 69 | }); 70 | 71 | // app.on('app:select-line', this.selectLines.bind(self)); 72 | var toggleBound = this.toggle.bind(this); 73 | app.on('app:loading', function() { $(".toggle-select-mode").off('click', toggleBound); }); 74 | app.on('app:loaded', function() { $(".toggle-select-mode").on('click', toggleBound); }); 75 | }; 76 | -------------------------------------------------------------------------------- /js/views/toolbar.js: -------------------------------------------------------------------------------- 1 | function Toolbar(opts) { 2 | for (let key in opts) { this[key] = opts[key]; } 3 | this.$el = $(this.el); 4 | } 5 | /** 6 | * Increase image zoom by Settings.zoomInFactor 7 | */ 8 | Toolbar.prototype.zoomIn = function(e) { 9 | e.stopPropagation(); 10 | $('#file-correction img').each(function() { 11 | Utils.scaleHeight(this, window.app.settings.zoomInFactor); 12 | }); 13 | }; 14 | 15 | /** 16 | * Decrease image zoom by Settings.zoomOutFactor 17 | */ 18 | Toolbar.prototype.zoomOut = function zoomOut(e) { 19 | e.stopPropagation(); 20 | $('#file-correction img').each(function() { 21 | Utils.scaleHeight(this, window.app.settings.zoomOutFactor); 22 | }); 23 | }; 24 | 25 | /** 26 | * Reset all images to their original size 27 | */ 28 | Toolbar.prototype.zoomReset = function zoomReset(e) { 29 | e.stopPropagation(); 30 | $('#file-correction img').each(function() { 31 | Utils.scaleHeight(this, 1); 32 | }); 33 | }; 34 | 35 | 36 | Toolbar.prototype.render = function() { 37 | var self = this; 38 | var app = window.app; 39 | 40 | // Save current page 41 | $("#save_button").on("click", app.savePage.bind(app)); 42 | 43 | // Open history modal 44 | $('button[data-target="#history-modal"]').on('click', app.showHistory.bind(app)); 45 | 46 | // Handle zooming 47 | $("#zoom-in").on("click", this.zoomIn); 48 | $("#zoom-out").on("click", this.zoomOut); 49 | $("#zoom-reset").on("click", this.zoomReset); 50 | 51 | // Handle view filtering by selectors 52 | this.$el.find(".set-view").on('click', function reduceView() { 53 | $(".view-hidden").removeClass("view-hidden"); 54 | $("ul.list-group > *").addClass('view-hidden'); 55 | $("ul.list-group > " + $(this).attr('data-target')).removeClass('view-hidden'); 56 | app.emit('app:filter-view'); 57 | }); 58 | 59 | // Handle sorting 60 | $("#sort-line").on('click', function() { app.pageView.sortRowsByLine(1); }); 61 | $("#sort-line-desc").on('click', function() { app.pageView.sortRowsByLine(-1); }); 62 | $("#sort-width").on('click', function() { app.pageView.sortRowsByWidth(1); }); 63 | $("#sort-width-desc").on('click', function() { app.pageView.sortRowsByWidth(-1); }); 64 | 65 | // 66 | // React to events 67 | // 68 | app.on('app:changed', function enableSaveButton() { 69 | $("#save_button").removeClass("disabled"); 70 | }); 71 | 72 | app.on('app:loaded', function disableSaveButton() { 73 | self.$el.find(".disabled").removeClass('disabled'); 74 | $("#save_button").addClass("disabled"); 75 | }); 76 | 77 | app.on('app:saving', function startSaveSpinner() { 78 | $("#wait_save").addClass("wait").removeClass("hidden"); 79 | $("#disk").addClass("hidden"); 80 | }); 81 | 82 | app.on('app:saved', function stopSaveSpinner() { 83 | $("#wait_save").removeClass("wait").addClass("hidden"); 84 | $("#disk").removeClass("hidden"); 85 | $("#save_button").addClass("disabled"); 86 | }); 87 | }; 88 | -------------------------------------------------------------------------------- /.jscsrc: -------------------------------------------------------------------------------- 1 | { 2 | "requireCurlyBraces": [ 3 | "for", 4 | "while", 5 | "do", 6 | "try", 7 | "catch" 8 | ], 9 | "requireOperatorBeforeLineBreak": true, 10 | "requireCamelCaseOrUpperCaseIdentifiers": { 11 | "allowedPrefixes": ["opt_", "GM_"], 12 | "allExcept": ["var_args"] 13 | }, 14 | "maximumLineLength": { 15 | "value": 200, 16 | "allExcept": ["comments", "regex"] 17 | }, 18 | "validateIndentation": 4, 19 | "validateQuoteMarks": false, 20 | 21 | "disallowMultipleLineStrings": true, 22 | "disallowMixedSpacesAndTabs": true, 23 | "disallowTrailingWhitespace": true, 24 | "disallowSpaceAfterPrefixUnaryOperators": true, 25 | "disallowMultipleVarDecl": true, 26 | "disallowKeywordsOnNewLine": ["else"], 27 | 28 | "requireSpaceAfterKeywords": [ 29 | "if", 30 | "else", 31 | "for", 32 | "while", 33 | "do", 34 | "switch", 35 | "return", 36 | "try", 37 | "catch" 38 | ], 39 | "requireSpaceBeforeBinaryOperators": [ 40 | "=", "+=", "-=", "*=", "/=", "%=", "<<=", ">>=", ">>>=", 41 | "&=", "|=", "^=", "+=", 42 | 43 | "+", "-", "*", "/", "%", "<<", ">>", ">>>", "&", 44 | "|", "^", "&&", "||", "===", "==", ">=", 45 | "<=", "<", ">", "!=", "!==" 46 | ], 47 | "requireSpaceAfterBinaryOperators": true, 48 | "requireSpacesInConditionalExpression": true, 49 | "requireSpaceBeforeBlockStatements": true, 50 | "requireSpacesInForStatement": true, 51 | "requireLineFeedAtFileEnd": true, 52 | "requireSpacesInFunctionExpression": { 53 | "beforeOpeningCurlyBrace": true 54 | }, 55 | "disallowSpacesInAnonymousFunctionExpression": { 56 | "beforeOpeningRoundBrace": true 57 | }, 58 | "disallowSpacesInsideObjectBrackets": "all", 59 | "disallowSpacesInsideArrayBrackets": "all", 60 | "disallowSpacesInsideParentheses": true, 61 | 62 | "disallowMultipleLineBreaks": false, 63 | "disallowNewlineBeforeBlockStatements": true, 64 | "disallowKeywords": ["with"], 65 | "disallowSpacesInFunctionExpression": { 66 | "beforeOpeningRoundBrace": true 67 | }, 68 | "disallowSpacesInFunctionDeclaration": { 69 | "beforeOpeningRoundBrace": true 70 | }, 71 | "disallowSpacesInCallExpression": true, 72 | "disallowSpaceAfterObjectKeys": true, 73 | "requireSpaceBeforeObjectValues": true, 74 | "requireCapitalizedConstructors": true, 75 | "requireDotNotation": true, 76 | "requireSemicolons": true, 77 | "validateParameterSeparator": ", ", 78 | "requireSpaceBetweenArguments": true, 79 | 80 | "jsDoc": { 81 | "checkAnnotations": "closurecompiler", 82 | "checkParamNames": true, 83 | "requireParamTypes": true, 84 | "checkRedundantParams": true, 85 | "checkReturnTypes": true, 86 | "checkRedundantReturns": true, 87 | "requireReturnTypes": true, 88 | "checkTypes": true, 89 | "checkRedundantAccess": true, 90 | "requireNewlineAfterDescription": true 91 | } 92 | } 93 | -------------------------------------------------------------------------------- /js/views/line-view.js: -------------------------------------------------------------------------------- 1 | function LineView(opts) { 2 | for (let key in opts) { this[key] = opts[key]; } 3 | this.tpl = window.app.templates.line; 4 | window.app.once('app:loaded', this.render.bind(this)); 5 | window.app.on('app:filter-view', this.renderToggler.bind(this)); 6 | window.app.on('app:enter-select-mode', this.renderCheckbox.bind(this)); 7 | window.app.on('app:exit-select-mode', this.renderCheckbox.bind(this)); 8 | } 9 | 10 | /* 11 | * Render (or do not render) the checkbox for multi-select mode 12 | */ 13 | LineView.prototype.renderCheckbox = function renderComment() { 14 | var self = this; 15 | // Selectionmode 16 | this.$el.find('.select-col').toggleClass('hidden', !window.app.selectMode); 17 | this.$el.find('.button-col').toggleClass('hidden', window.app.selectMode); 18 | this.$el.toggleClass('selected', this.selected); 19 | this.$el.find(':checkbox').prop('checked', this.selected); 20 | }; 21 | 22 | LineView.prototype.renderTextarea = function renderTextarea() { 23 | // fit height 24 | Utils.fitHeight(this.$el.find('textarea')); 25 | }; 26 | 27 | /** 28 | * Update the color of the comment toggle button depending on whether line has 29 | * comments or not. 30 | */ 31 | LineView.prototype.renderToggler = function renderToggler() { 32 | var lineComment = this.$el.find('.line-comment'); 33 | var isVisible = lineComment.is(':visible'); 34 | var hasComment = this.model.comment.length > 0; 35 | var $toggler = this.$el.find(".toggle-line-comment"); 36 | $toggler.find(".show-line-comment").toggleClass('hidden', isVisible); 37 | $toggler.find(".hide-line-comment").toggleClass('hidden', !isVisible); 38 | $toggler.toggleClass('btn-default', !hasComment).toggleClass('btn-info', hasComment); 39 | }; 40 | 41 | LineView.prototype.addTag = function addTag(tag) { 42 | this.model.addTag(tag); 43 | this.render(); 44 | }; 45 | 46 | LineView.prototype.onInput = function onInput() { 47 | this.model.comment = this.$el.find('.line-comment textarea ').val().trim(); 48 | this.model.transcription = this.$el.find('.line-transcription input').val().trim(); 49 | this.renderToggler(); 50 | this.renderTextarea(); 51 | window.app.emit('app:changed'); 52 | }; 53 | 54 | LineView.prototype.render = function() { 55 | var self = this; 56 | console.log("Rendering", this.model.id); 57 | 58 | // Build from template 59 | this.$el.off().find("*").off(); 60 | this.$el.html($(self.tpl(this.model))); 61 | 62 | // data binding 63 | this.$el.on('input', self.onInput.bind(this)); 64 | 65 | // Mark line selected on click in select mode 66 | this.$el.on('click', function() { 67 | if (window.app.selectMode) { 68 | self.selected = !self.selected; 69 | self.renderCheckbox(); 70 | } 71 | }); 72 | 73 | // Add error tag on click 74 | this.$el.find("*[data-tag]").on('click', function() { 75 | self.addTag($(this).attr('data-tag')); 76 | window.app.emit('app:changed'); 77 | }); 78 | 79 | // On clicking the comment toggler 80 | this.$el.find(".toggle-line-comment").on('click', function() { 81 | var commentField = self.$el.find('.line-comment'); 82 | commentField.toggleClass('hidden', commentField.is(':visible')).removeClass('view-hidden'); 83 | self.renderToggler(); 84 | }); 85 | 86 | // Render the toggle button 87 | this.renderToggler(); 88 | 89 | // Render (or don't) the checkbox 90 | this.renderCheckbox(); 91 | 92 | // Fit height of text area 93 | this.renderTextarea(); 94 | 95 | return this; 96 | }; 97 | -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | ocr-gt-tools 2 | ============ 3 | 4 | A web interface for creating ground truth for evaluating and training OCR. 5 | 6 | [](https://hub.docker.com/r/ubma/ocr-gt-tools/) 7 | [](https://hub.docker.com/r/ubma/ocr-gt-tools/) 8 | [](https://github.com/UB-Mannheim/ocr-gt-tools/blob/master/LICENSE) 9 | [](https://github.com/UB-Mannheim/ocr-gt-tools/issues) 10 | [](https://travis-ci.org/UB-Mannheim/ocr-gt-tools) 11 | [](https://github.com/UB-Mannheim/ocr-gt-tools/stargazers) 12 | 13 | ## Table of Contents 14 | 15 | * [Summary](#summary) 16 | * [Screenshot](#screenshot) 17 | * [Features](#features) 18 | * [Installation](#installation) 19 | * [About the code](#about-the-code) 20 | * [Usage](#usage) 21 | * [Contributing](#contributing) 22 | * [Pull Requests](#pull-requests) 23 | * [Bug reports](#bug-reports) 24 | * [Acknowledgments](#acknowledgments) 25 | 26 | ## Summary 27 | 28 | ocr-gt-tools allows editing 29 | [hOCR](https://github.com/kba/hocr-spec/blob/master/hocr-spec.md) files, 30 | such as those produced by the 31 | [tesseract](https://github.com/tesseract-ocr/tesseract) or 32 | [ocropy](https://github.com/tmbdev/ocropy) OCR frameworks. 33 | 34 | ### Screenshot 35 | 36 |
37 |
38 | ### Features
39 |
40 | * Editing transcriptions of lines
41 | * Commenting on line and page level
42 | * Use [standardized comment tags](https://github.com/UB-Mannheim/ocr-gt-tools/wiki/Error-Tags) to mark common problems
43 | * [Cheatsheet](./doc/screenshots/cheatsheet-2016-05-04.png)
44 | * Zoom in / Zoom out
45 | * Filter visible elements
46 | * Select multiple lines and apply tags.
47 |
48 | ### Installation
49 |
50 | See [INSTALL.md](./INSTALL.md).
51 |
52 | ### About the code
53 |
54 | The server-side code is written in Perl.
55 |
56 | The frontend is written in HTML and Javascript.
57 |
58 | ## Usage
59 |
60 | - Open 'ocr-gt-tools/index.html' with a browser
61 | - open in a second Window 'Page Previews' from Kitodo
62 | - Search the book from which you created the hOCR file
63 | - Drag and drop a image from the Kitodo 'Page Preview' Window to the Window with 'ocr-gt-tools/index.html'
64 | - The perl script ocr-gt-tools.cgi will create in the background all files, which takes a few seconds
65 | - with ajax a json objects will be returned to index.html
66 | - index.html will load with ajax the created 'correction.html' and 'anmerkungen.txt' inline
67 | - 'Speichern' will get active if you have written a comment or a text line
68 |
69 | ## Contributing
70 |
71 | ### Expand the wiki
72 |
73 | We are using the wiki to collect [transcription hints for unusual
74 | glyphs](wiki/Special-Characters) and [frequent errors](wiki/Error-Tags).
75 |
76 | ### Pull Requests
77 |
78 | Bug fixes, new functions, suggestions for new features and other user feedback
79 | are appreciated.
80 |
81 | The source code is available from https://github.com/UB-Mannheim/ocr-gt-tools.
82 | Please prepare your code contributions also on Github.
83 |
84 | ### Bug reports
85 |
86 | Please feel free to [open
87 | issues](https://github.com/UB-Mannheim/ocr-gt-tools/issues) for any bug you
88 | encounter and features you'd like to have.
89 |
90 |
91 | ## Acknowledgments
92 |
93 | This is free software. You may use it under the terms of the
94 | GNU AFFERO General Public License (AGPL) version 3 or newer.
95 | See [LICENSE](LICENSE) for details.
96 |
97 | This project bundles other free software:
98 |
99 | * [EB Garamond Font](https://www.google.com/fonts/specimen/EB+Garamond) (SIL Open Font License)
100 | * [Font Awesome by Dave Gandy](http://fontawesome.io/) (SIL OFL 1.1, MIT)
101 | * [bootstrap](http://getbootstrap.com/) (MIT)
102 | * [clipboard.js](https://github.com/zenorocha/clipboard.js) (MIT)
103 | * [handlebars.js](https://github.com/wycats/handlebars.js) (MIT)
104 | * [hocr-extract-images](https://github.com/tmbdev/hocr-tools) (Apache)
105 | * [jQuery](http://jquery.com/) (MIT)
106 | * [ocropus-gtedit](https://github.com/tmbdev/ocropy) (Apache)
107 | * [reset-css](https://github.com/shannonmoeller/reset-css) (Public Domain)
108 |
--------------------------------------------------------------------------------
/ocr-gt-tools.styl:
--------------------------------------------------------------------------------
1 | selectedColor = red
2 | paperColor = #b9af96
3 |
4 | @media (min-width: 768px)
5 | body
6 | padding-top 0
7 | .container-fluid
8 | margin-left 100px
9 | .navbar-collapse
10 | height auto
11 | border-top 0
12 | box-shadow none
13 | max-height none
14 | padding-left 0
15 | padding-right 0
16 | &.collapse
17 | display block !important
18 | width auto !important
19 | padding-bottom 0
20 | overflow visible !important
21 | &.in
22 | overflow-x visible
23 | .navbar-nav
24 | &.navbar-right
25 | &:last-child
26 | margin-right 0
27 | .navbar
28 | max-width 96px
29 | height 100vh
30 | margin-right 0
31 | margin-left 0
32 | float left
33 | position fixed
34 | z-index 10001
35 | &:after
36 | clear both
37 | .btn-group
38 | display block
39 | .dropdown-menu
40 | top 0
41 | left 100%
42 | padding 0
43 | width 220px
44 | .btn.disabled
45 | opacity .35
46 | .navbar-nav,
47 | .navbar-nav > li,
48 | .navbar-left,
49 | .navbar-right,
50 | .navbar-header
51 | float none !important
52 | .navbar-right
53 | .dropdown-menu
54 | left 0
55 | right auto
56 | .navbar-default
57 | .dropdown-menu
58 | i
59 | display inline-block
60 | float left
61 | top 2px
62 | font-size 2em
63 | .modal-admin
64 | width 80%
65 | left 100px
66 | position fixed
67 | max-height 100vh
68 | overflow scroll
69 | .line
70 | .list-group-item
71 | padding 0 0 0 0
72 | .btn
73 | i
74 | padding-top 0
75 | padding-bottom 0
76 | font-size: 20px
77 |
78 | #dropzone
79 | position fixed
80 | height 90vh
81 | margin 5vh
82 | padding-top 40vh
83 | font-size 300%
84 | text-align center
85 | &.droppable
86 | border-style dashed
87 | border-width 10px
88 | border-color blue
89 |
90 | #waiting-animation
91 | position: fixed
92 | height: 100vh
93 | width: 100vw
94 | z-index: 10000
95 | a
96 | position: absolute
97 | height: 32px
98 | width: 32px
99 | img
100 | height: 32px !important
101 | max-width: 32px !important
102 | display: block !important
103 |
104 | .hidden
105 | display none
106 |
107 | .view-hidden
108 | display none
109 |
110 | .selected
111 | background: selectedColor
112 | .list-group-item.image
113 | background: selectedColor
114 | .panel
115 | background: selectedColor
116 |
117 | textarea
118 | input[type='text']
119 | display inline-block
120 | white-space pre-wrap
121 | border none
122 | padding 0
123 | color #000060
124 | font-family 'EB Garamond', serif
125 | font-size 20px
126 | min-height 24px
127 | width 100%
128 | // Important to make textareas auto-expand
129 | height auto
130 | overflow hidden
131 | resize none
132 |
133 | .line-comment, .line-comment *,
134 | #page-comment textarea
135 | background-color lightyellow
136 | font-family 'EB Garamond', serif
137 | font-size 20px
138 | min-height 24px
139 |
140 | #file-correction
141 | .panel
142 | margin 0
143 | .panel-heading
144 | padding 0
145 | h4
146 | padding-top 7.5px
147 | .col-sm-1
148 | width:10%
149 | padding-left 0
150 | padding-right 0
151 | .btn-group > .btn
152 | padding: 0
153 | width: 35px
154 | .btn
155 | text-align left
156 | i
157 | padding: 5px;
158 | .col-sm-11
159 | width: 88%
160 | padding-left: 0
161 |
162 |
163 | #right-sidebar
164 | position fixed
165 | right 0
166 | .list-group-item
167 | i
168 | padding-right: 5px;
169 |
170 | #select-bar
171 | position fixed
172 | background-color #eee
173 | border-radius 0 0 10px 10px
174 | border-left: 2px solid blue
175 | border-right 2px solid blue
176 | border-bottom 2px solid blue
177 | top 0
178 | left 200px
179 | z-index: 1001
180 | .close
181 | opacity .5
182 | display inline-block
183 | padding-right 5px
184 | font-size xx-large
185 |
186 | #cheatsheet-modal
187 | // disable selection
188 | -webkit-touch-callout: none;
189 | -webkit-user-select: none;
190 | -khtml-user-select: none;
191 | -moz-user-select: none;
192 | -ms-user-select: none;
193 | user-select: none;
194 | h4
195 | float: left
196 | .cheatsheet-entry
197 | float: left
198 | margin-left 25%
199 | font-size: 200%
200 | max-width: 64px
201 | img
202 | height: 64px
203 | th
204 | font-weight: bold
205 | td:nth-child(1)
206 | width: 40%
207 | td
208 | vertical-align: middle
209 | font-size: 120%
210 | button.code
211 | width 100%
212 | font-family monospace
213 | font-size 48px
214 | height 64px
215 | .clipboard
216 | float: left
217 | font-size: 50%
218 |
219 | .select-col
220 | padding-left: 20px
221 | // margin-top: 32px
222 | input[type="checkbox"]
223 | width 23px
224 | height 23px
225 | padding: 0
226 | margin: 4px 0 0 0
227 | // padding-top 32px
228 |
229 |
230 | #work-info
231 | overflow-y: scroll
232 | max-height: 200px;
233 |
234 | // vim: sw=4 ts=4 noet :
235 |
--------------------------------------------------------------------------------
/Makefile:
--------------------------------------------------------------------------------
1 | # Add node_modules/.bin to $PATH so the CLI tools
2 | # installed locally by npm can be used
3 | export PATH := $(PWD)/node_modules/.bin:$(PATH)
4 |
5 | include \
6 | dev/apache.mk \
7 | dev/debian.mk \
8 | dev/docker.mk \
9 | dev/plackup.mk
10 | #
11 | # Define all the CLI tools to use
12 | #
13 |
14 | # Standard UNIX tools, recurse, create parents, force delete
15 | MKDIR = mkdir -p
16 | RM = rm -rf
17 | CP = cp -r
18 | GIT_CLONE = git clone --depth 1
19 |
20 | # cURL to download files
21 | CURL = curl -s
22 | # clean-css is a CSS minifier and optimizer
23 | CLEANCSS = cleancss
24 | # UglifyJS minifies, merges and optimizes Javascript
25 | UGLIFYJS = uglifyjs
26 | # webfont-dl is a tool to download web fonts from the Google Fonts API
27 | WEBFONTDL = webfont-dl --eot=omit --ttf=data --woff1=data
28 | # Pug is a templating engine
29 | PUG = pug --pretty
30 | # Stylus is a CSS compiler
31 | STYLUS = stylus
32 |
33 | #
34 | # Define lists of assets
35 | #
36 |
37 | # URLs of Web Fonts to embed
38 | FONT_URLS = https://fonts.googleapis.com/css?family=EB+Garamond&subset=latin,latin-ext
39 | # Font files (eot, ttf, woff...) to bundle
40 | FONT_FILES = node_modules/font-awesome/fonts/fontawesome-webfont.* \
41 | node_modules/bootstrap/fonts/glyphicons-halflings-regular.*
42 | # URLs of CSS to download
43 | CSS_URLS = https://getbootstrap.com/docs/3.4/examples/dashboard/dashboard.css
44 | # CSS files to bundle into one minified `dist/vendor.css`
45 | # NOTE: Our CSS should not be bundled here
46 | CSS_FILES = node_modules/reset-css/reset.css \
47 | node_modules/bootstrap/dist/css/bootstrap.css \
48 | node_modules/notie/dist/notie.css \
49 | node_modules/font-awesome/css/font-awesome.css
50 | # JS scripts to bundle into one minified `dist/vendor.js`
51 | # NOTE: Javascript developed by us should not be bundled here
52 | VENDOR_JS_FILES = node_modules/jquery/dist/jquery.js \
53 | node_modules/async/dist/async.min.js \
54 | node_modules/bootstrap/dist/js/bootstrap.js \
55 | node_modules/handlebars/dist/handlebars.min.js \
56 | node_modules/clipboard/dist/clipboard.js \
57 | node_modules/notie/dist/notie.js
58 |
59 | JS_FILES = js/*.js js/**/*.js ocr-gt-tools.js ocr-gt-tools.js
60 | # The HTML files, described in the Pug shorthand / templating language
61 | PUG_FILES = ocr-gt-tools.pug
62 |
63 | #
64 | # Define the list of targets that will "always fail", i.e. the CLI api
65 | #
66 | # clean-js clean-html clean-fonts clean-css \
67 |
68 | .PHONY: debug clean vendor
69 |
70 | #
71 | # Debugging
72 | #
73 | print-%: ; @echo $*=$($*)
74 |
75 | __: clean dist
76 |
77 | _.%: ; $(MAKE) -C . clean-$* dist
78 |
79 | debug:
80 | @grep '^[A-Z0-9_]\+\s*=' Makefile \
81 | |grep -o '^[A-Z0-9_]*' \
82 | |xargs -I{} make -s . print-{}
83 |
84 | #
85 | # Dependencies to execute ocropy / hocr-tools in a CGI environment
86 | #
87 |
88 | vendor: dist/vendor/hocr-tools dist/vendor/ocropy
89 |
90 | dist/vendor/hocr-tools:
91 | $(MKDIR) dist/vendor
92 | $(GIT_CLONE) https://github.com/UB-Mannheim/hocr-tools $@
93 |
94 | dist/vendor/ocropy:
95 | $(MKDIR) dist/vendor
96 | $(GIT_CLONE) https://github.com/tmbdev/ocropy $@
97 |
98 | #
99 | # Set up dist folder
100 | #
101 |
102 | dist: \
103 | dist/special-chars.json\
104 | dist/error-tags.json\
105 | dist/vendor\
106 | dist/log\
107 | dist/vendor.css\
108 | dist/vendor.js\
109 | dist/fonts\
110 | dist/index.html\
111 | dist/ocr-gt-tools.js\
112 | dist/ocr-gt-tools.css\
113 | dist/ocr-gt-tools.cgi
114 |
115 | dist/%.json: doc/%.json
116 | $(CP) $< $@
117 |
118 | dist/log:
119 | $(MKDIR) $@
120 |
121 | dist/ocr-gt-tools.cgi: ocr-gt-tools.cgi
122 | $(CP) $< $@
123 | chmod a+x $@
124 |
125 | #$(UGLIFYJS) --compress --output $@ $^
126 | dist/ocr-gt-tools.js: $(JS_FILES)
127 | cat $^ > $@
128 |
129 | dist/ocr-gt-tools.css: ocr-gt-tools.styl
130 | $(STYLUS) < $< > $@
131 |
132 | dist/fonts:
133 | $(MKDIR) $@
134 | $(CP) ${FONT_FILES} $@
135 |
136 | dist/fonts.css: dist/fonts
137 | $(WEBFONTDL) -o $@ --font-out=dist/fonts $(FONT_URLS) && wait
138 |
139 | dist/vendor.css: ${CSS_FILES} dist/fonts.css
140 | cat dist/fonts.css ${CSS_FILES} \
141 | | sed 's,\.\./fonts,fonts,g' \
142 | > dist/temp.css
143 | $(CURL) ${CSS_URLS} >> dist/temp.css
144 | $(CLEANCSS) --skip-rebase --output $@ dist/temp.css
145 | $(RM) dist/temp.css
146 |
147 | dist/vendor.js: ${VENDOR_JS_FILES}
148 | $(UGLIFYJS) --output $@ \
149 | --prefix 1 \
150 | --source-map $@.map \
151 | --source-map-url vendor.js.map \
152 | $^
153 |
154 | # sed "s,\(=.\)dist/,\1,g" $< | $(PUG) > $@
155 | dist/index.html: ${PUG_FILES}
156 | $(MKDIR) dist
157 | $(PUG) < $< > $@
158 |
159 | #
160 | # Clean up, delete files
161 | #
162 |
163 | clean-fonts:
164 | $(RM) dist/fonts dist/fonts.css
165 |
166 | clean-%:
167 | $(RM) dist/$* dist/*.$* dist/*.$*.map
168 |
169 | clean: clean-js clean-css clean-fonts clean-html
170 |
171 | realclean:
172 | $(RM) node_modules
173 | $(RM) dist
174 |
175 | test:
176 | bash ./test.sh
177 |
--------------------------------------------------------------------------------
/dist/error-tags.json:
--------------------------------------------------------------------------------
1 | {
2 | "wrong-image-section": {
3 | "name": {
4 | "en": "Incorrectly captured image section which contains no straight line of text but empty spaces or page margins etc.",
5 | "de": "Falsch erfasste Bildausschnitte, die keine Textzeile enthalten, sondern leere Seitenbereiche oder Seitenränder usw."
6 | },
7 | "total": 443,
8 | "frequencyAvg": 5.7,
9 | "id": "wrong-image-section"
10 | },
11 | "text-blocked": {
12 | "name": {
13 | "en": "Blocked text with visible blanks between the letters",
14 | "de": "Gesperrter Text mit sichtbaren Leerzeichen zwischen den Buchstaben"
15 | },
16 | "total": 51,
17 | "frequencyAvg": 0.7,
18 | "id": "text-blocked"
19 | },
20 | "text-italic": {
21 | "name": {
22 | "en": "Line completely or in parts in italics",
23 | "de": "Zeile vollständig oder teilweise in kursiver Schrift"
24 | },
25 | "total": 156,
26 | "frequencyAvg": 2,
27 | "id": "text-italic"
28 | },
29 | "initial": {
30 | "name": {
31 | "en": "initial character",
32 | "de": "Initiale"
33 | },
34 | "comment": {
35 | "en": "This most likely causes wrong-image-section"
36 | },
37 | "total": 10,
38 | "frequencyAvg": 0.1,
39 | "id": "initial"
40 | },
41 | "letter-faded": {
42 | "name": {
43 | "en": "Partially or completely faded letters",
44 | "de": "Teilweise oder vollständig ausgebleichte Buchstaben"
45 | },
46 | "total": 6,
47 | "frequencyAvg": 0.1,
48 | "id": "letter-faded"
49 | },
50 | "notes-within-line": {
51 | "name": {
52 | "en": "Notes on page margin captured within a text line",
53 | "de": "Anmerkungen am Seitenrand mit Textzeile erfasst"
54 | },
55 | "total": 93,
56 | "frequencyAvg": 1.2,
57 | "id": "notes-within-line"
58 | },
59 | "notes-separate": {
60 | "name": {
61 | "en": "Notes on page margin captured as separate lines",
62 | "de": "Anmerkungen am Seitenrand als separate Zeilen erfasst"
63 | },
64 | "total": 95,
65 | "frequencyAvg": 1.2,
66 | "id": "notes-separate"
67 | },
68 | "letter-handling-unclear": {
69 | "name": {
70 | "en": "Characters whose treatment is not yet clear",
71 | "de": "Buchstaben deren Behandlung noch nicht klar ist"
72 | },
73 | "comment": {
74 | "de": "Zum Beispiel q mit Akut, que-Ligatur"
75 | },
76 | "total": 17,
77 | "frequencyAvg": 0.2,
78 | "id": "letter-handling-unclear"
79 | },
80 | "line-incomplete": {
81 | "name": {
82 | "en": "line not captured completely",
83 | "de": "Zeile nicht vollständig erfasst"
84 | },
85 | "comment": {
86 | "de": "Zeile wurde zwar korrekt erfasst, aber Buchtsaben links oder rechts in der Zeile fehlen",
87 | "en": "Line was captured correctly, but letters in the left or right of the line are missing"
88 | },
89 | "total": 33,
90 | "frequencyAvg": 0.4,
91 | "id": "line-incomplete"
92 | },
93 | "line-incorrect": {
94 | "name": {
95 | "en": "line not captured correctly",
96 | "de": "Zeile nicht richtig erfasst"
97 | },
98 | "comment": {
99 | "en": "more than just one line inside the image; line lies at an angle in image",
100 | "de": "Mehrere Zeilen im Bild erfasst; Zeile liegt schräg im Bild"
101 | },
102 | "total": 57,
103 | "frequencyAvg": 0.7,
104 | "id": "line-incorrect"
105 | },
106 | "line-captured-twice": {
107 | "name": {
108 | "en": "line partially or completely captured tice",
109 | "de": "Zeile teilweise oder vollständig doppelt erfasst"
110 | },
111 | "total": 33,
112 | "frequencyAvg": 0.4,
113 | "id": "line-captured-twice"
114 | },
115 | "text-greek": {
116 | "name": {
117 | "en": "Greek text",
118 | "de": "Griechischer Text"
119 | },
120 | "total": 3,
121 | "frequencyAvg": 0,
122 | "id": "text-greek"
123 | },
124 | "letter-unidentified": {
125 | "name": {
126 | "en": "Letter not jet identified",
127 | "de": "Noch nicht genau identifizierter Buchstabe"
128 | },
129 | "comment": {
130 | "en": "May be similar with letter-handling-unclear"
131 | },
132 | "total": 12,
133 | "frequencyAvg": 0.2,
134 | "id": "letter-unidentified"
135 | },
136 | "letter-unreadable": {
137 | "name": {
138 | "en": "letter not faded but still unreadable",
139 | "de": "Buchstabe nicht ausgebleicht aber trotzdem unleserlich"
140 | },
141 | "total": 2,
142 | "frequencyAvg": 0,
143 | "id": "letter-unreadable"
144 | },
145 | "line-not-in-order": {
146 | "name": {
147 | "en": "line not captured in correct order",
148 | "de": "Zeile nicht in richtiger Reihenfolge erfasst"
149 | },
150 | "total": 4,
151 | "frequencyAvg": 0.1,
152 | "id": "line-not-in-order"
153 | },
154 | "dividing-line": {
155 | "name": {
156 | "en": "Dividing line captured as line",
157 | "de": "Trennlinie als Zeile erfasst"
158 | },
159 | "total": 2,
160 | "id": "dividing-line"
161 | }
162 | }
163 |
--------------------------------------------------------------------------------
/doc/error-tags.json:
--------------------------------------------------------------------------------
1 | {
2 | "wrong-image-section": {
3 | "name": {
4 | "en": "Incorrectly captured image section which contains no straight line of text but empty spaces or page margins etc.",
5 | "de": "Falsch erfasste Bildausschnitte, die keine Textzeile enthalten, sondern leere Seitenbereiche oder Seitenränder usw."
6 | },
7 | "total": 443,
8 | "frequencyAvg": 5.7,
9 | "id": "wrong-image-section"
10 | },
11 | "text-blocked": {
12 | "name": {
13 | "en": "Blocked text with visible blanks between the letters",
14 | "de": "Gesperrter Text mit sichtbaren Leerzeichen zwischen den Buchstaben"
15 | },
16 | "total": 51,
17 | "frequencyAvg": 0.7,
18 | "id": "text-blocked"
19 | },
20 | "text-italic": {
21 | "name": {
22 | "en": "Line completely or in parts in italics",
23 | "de": "Zeile vollständig oder teilweise in kursiver Schrift"
24 | },
25 | "total": 156,
26 | "frequencyAvg": 2,
27 | "id": "text-italic"
28 | },
29 | "initial": {
30 | "name": {
31 | "en": "initial character",
32 | "de": "Initiale"
33 | },
34 | "comment": {
35 | "en": "This most likely causes wrong-image-section"
36 | },
37 | "total": 10,
38 | "frequencyAvg": 0.1,
39 | "id": "initial"
40 | },
41 | "letter-faded": {
42 | "name": {
43 | "en": "Partially or completely faded letters",
44 | "de": "Teilweise oder vollständig ausgebleichte Buchstaben"
45 | },
46 | "total": 6,
47 | "frequencyAvg": 0.1,
48 | "id": "letter-faded"
49 | },
50 | "notes-within-line": {
51 | "name": {
52 | "en": "Notes on page margin captured within a text line",
53 | "de": "Anmerkungen am Seitenrand mit Textzeile erfasst"
54 | },
55 | "total": 93,
56 | "frequencyAvg": 1.2,
57 | "id": "notes-within-line"
58 | },
59 | "notes-separate": {
60 | "name": {
61 | "en": "Notes on page margin captured as separate lines",
62 | "de": "Anmerkungen am Seitenrand als separate Zeilen erfasst"
63 | },
64 | "total": 95,
65 | "frequencyAvg": 1.2,
66 | "id": "notes-separate"
67 | },
68 | "letter-handling-unclear": {
69 | "name": {
70 | "en": "Characters whose treatment is not yet clear",
71 | "de": "Buchstaben deren Behandlung noch nicht klar ist"
72 | },
73 | "comment": {
74 | "de": "Zum Beispiel q mit Akut, que-Ligatur"
75 | },
76 | "total": 17,
77 | "frequencyAvg": 0.2,
78 | "id": "letter-handling-unclear"
79 | },
80 | "line-incomplete": {
81 | "name": {
82 | "en": "line not captured completely",
83 | "de": "Zeile nicht vollständig erfasst"
84 | },
85 | "comment": {
86 | "de": "Zeile wurde zwar korrekt erfasst, aber Buchtsaben links oder rechts in der Zeile fehlen",
87 | "en": "Line was captured correctly, but letters in the left or right of the line are missing"
88 | },
89 | "total": 33,
90 | "frequencyAvg": 0.4,
91 | "id": "line-incomplete"
92 | },
93 | "line-incorrect": {
94 | "name": {
95 | "en": "line not captured correctly",
96 | "de": "Zeile nicht richtig erfasst"
97 | },
98 | "comment": {
99 | "en": "more than just one line inside the image; line lies at an angle in image",
100 | "de": "Mehrere Zeilen im Bild erfasst; Zeile liegt schräg im Bild"
101 | },
102 | "total": 57,
103 | "frequencyAvg": 0.7,
104 | "id": "line-incorrect"
105 | },
106 | "line-captured-twice": {
107 | "name": {
108 | "en": "line partially or completely captured tice",
109 | "de": "Zeile teilweise oder vollständig doppelt erfasst"
110 | },
111 | "total": 33,
112 | "frequencyAvg": 0.4,
113 | "id": "line-captured-twice"
114 | },
115 | "text-greek": {
116 | "name": {
117 | "en": "Greek text",
118 | "de": "Griechischer Text"
119 | },
120 | "total": 3,
121 | "frequencyAvg": 0,
122 | "id": "text-greek"
123 | },
124 | "letter-unidentified": {
125 | "name": {
126 | "en": "Letter not jet identified",
127 | "de": "Noch nicht genau identifizierter Buchstabe"
128 | },
129 | "comment": {
130 | "en": "May be similar with letter-handling-unclear"
131 | },
132 | "total": 12,
133 | "frequencyAvg": 0.2,
134 | "id": "letter-unidentified"
135 | },
136 | "letter-unreadable": {
137 | "name": {
138 | "en": "letter not faded but still unreadable",
139 | "de": "Buchstabe nicht ausgebleicht aber trotzdem unleserlich"
140 | },
141 | "total": 2,
142 | "frequencyAvg": 0,
143 | "id": "letter-unreadable"
144 | },
145 | "line-not-in-order": {
146 | "name": {
147 | "en": "line not captured in correct order",
148 | "de": "Zeile nicht in richtiger Reihenfolge erfasst"
149 | },
150 | "total": 4,
151 | "frequencyAvg": 0.1,
152 | "id": "line-not-in-order"
153 | },
154 | "dividing-line": {
155 | "name": {
156 | "en": "Dividing line captured as line",
157 | "de": "Trennlinie als Zeile erfasst"
158 | },
159 | "total": 2,
160 | "id": "dividing-line"
161 | }
162 | }
163 |
--------------------------------------------------------------------------------
/INSTALL.md:
--------------------------------------------------------------------------------
1 | # Installation Instructions
2 |
3 | * [Docker Quickstart](#docker-quickstart)
4 | * [Install dependencies](#install-dependencies)
5 | * [Create configuration](#create-configuration)
6 | * [Deploy on a server](#deploy-on-a-server)
7 | * [On Apache](#on-apache)
8 | * [Bundled standalone server](#bundled-standalone-server)
9 | * [Testing the server](#testing-the-server)
10 | * [Developing the frontend](#developing-the-frontend)
11 | * [Perl](#perl)
12 | * [Log-Files / Error-Log-Files](#log-files--error-log-files)
13 |
14 | ## Docker Quickstart
15 |
16 | To get the tool up and running in a docker container:
17 |
18 | ```
19 | git clone https://github.com/UB-Mannheim/ocr-gt-tools
20 | cd ocr-gt-tools
21 | ./dev/run-docker.sh ");
50 |
51 | // Set up models
52 | this.settings = new Settings();
53 | this.history = new History();
54 | this.errorTags = new ErrorTags();
55 | this.cheatsheet = new Cheatsheet();
56 |
57 | // Compile templates
58 | this.templates = Utils.compileTemplates();
59 | }
60 |
61 | App.prototype.on = function() {
62 | if (this.settings.debug) console.info('bind', arguments[0], ' -> ', arguments[1].name || arguments[1]);
63 | this.$el.on.apply(this.$el, arguments);
64 | return this;
65 | };
66 | App.prototype.once = function() {
67 | if (this.settings.debug) console.info('bind', arguments[0], ' -> ', arguments[1].name || arguments[1]);
68 | this.$el.one.apply(this.$el, arguments);
69 | return this;
70 | };
71 |
72 |
73 | App.prototype.emit = function() {
74 | if (this.settings.debug) console.log('emit', arguments[0], arguments[1] || '');
75 | var event = arguments[0];
76 | if (event === 'app:saved') {
77 | notie.alert(1, "Gespeichert", 1);
78 | } else if (event === 'app:ajaxError') {
79 | var xhr = arguments[1];
80 | notie.alert(3, "HTTP Fehler " + xhr.status + ":\n
163 |
169 | `);
170 | $("#glyph-input").on('keyup', function(e) {
171 | var $input = $("#glyph-input");
172 | var from = $input[0].selectionStart;
173 | var to = $input[0].selectionEnd;
174 | if (from == to) {
175 | from -= 1;
176 | }
177 | $('#glyph-propose').empty();
178 | var $propose = $('#glyph-propose');
179 | var val = $input.val();
180 | var chosen = val.substring(from, to);
181 | console.log(chosen, from, to);
182 | $.each(scraped, function() {
183 | var glyphDesc = this;
184 | if (glyphDesc.baseLetter.indexOf(chosen) === -1) {
185 | return;
186 | }
187 | $.each(glyphDesc.sample, function(i, sample) {
188 | $propose.append($(sample)
189 | .on('click', function(e) {
190 | e.preventDefault();
191 | $input.val(val.substr(0, from) + glyphDesc.recognition + val.substr(to));
192 | }));
193 | });
194 | });
195 | });
196 | },
197 | 'Error-Tags': function(scraped) {
198 | console.log('Not Implemented for Error-Tags (yet?)');
199 | }
200 | };
201 |
202 | function scrapeJsonFromWikiPage(schema) {
203 | var parsed = {};
204 | var validator = new ZSchema();
205 | var h2s = $(".markdown-body h2").get();
206 | for (let i = 0; i < h2s.length; i++) {
207 | var $h2 = $(h2s[i]);
208 | var thingDesc = {};
209 | var thingId = $h2.text().trim();
210 | parsed[thingId] = thingDesc;
211 | var lis = $h2.next('ul').find('li').get();
212 | for (let j = 0; j < lis.length; j++) {
213 | var liHtml = $(lis[j]).html();
214 | var colonIndex = liHtml.indexOf(':');
215 | var varName = liHtml.substring(0, colonIndex)
216 | .toLowerCase()
217 | .replace(/[^a-z0-9]+/g, '_')
218 | .replace(/_([a-z])/g, function(orig, ch) {
219 | return ch.toUpperCase();
220 | })
221 | .replace(/^_|_$/, '');
222 | // console.log(`Parsing '${varName}'`);
223 | var rawValue = liHtml.substring(colonIndex + 1).trim();
224 | rawValue = rawValue.replace('&', '&');
225 | if (schema.properties[varName] && schema.properties[varName].type === 'array') {
226 | thingDesc[varName] = rawValue.split(/\s*;\s*/);
227 | } else if (schema.properties[varName] && schema.properties[varName].type === 'number') {
228 | thingDesc[varName] = parseFloat(rawValue);
229 | } else if (/[A-Z][a-z]$/.test(varName)) {
230 | var lang = varName.substr(-2).toLowerCase();
231 | varName = varName.substring(0, varName.length - 2);
232 | thingDesc[varName] = thingDesc[varName] || {};
233 | thingDesc[varName][lang] = rawValue;
234 | } else {
235 | thingDesc[varName] = rawValue;
236 | }
237 | }
238 | thingDesc.id = thingId;
239 | console.log([thingDesc, schema]);
240 | if (!validator.validate(thingId, schema.properties.id)) {
241 | showError(thingId, validator.getLastErrors());
242 | }
243 | if (!validator.validate(thingDesc, schema)) {
244 | showError(thingId, validator.getLastErrors());
245 | }
246 | }
247 | return parsed;
248 | };
249 |
250 | function escapeHTML(str) {
251 | var entityMap = {
252 | "&": "&",
253 | "<": "<",
254 | ">": ">",
255 | '"': '"',
256 | "'": ''',
257 | "/": '/'
258 | };
259 | return String(str).replace(/[&<>"'\/]/g, function(s) {
260 | return entityMap[s];
261 | });
262 | }
263 |
264 | function showError(faultyId, err) {
265 | $(`h2:contains('${faultyId}')`).append(
266 | `" + xhr.responseText + ""); 81 | } 82 | this.$el.trigger.apply(this.$el, arguments); 83 | return this; 84 | }; 85 | 86 | App.prototype.confirmExit = function confirmExit() { 87 | if (this.currentPage && this.currentPage.changed) { 88 | notie.alert(2, "Ungesicherte Inhalte vorhanden, bitte zuerst speichern!", 5); 89 | return "Ungesicherte Inhalte vorhanden, bitte zuerst speichern!"; 90 | } 91 | }; 92 | 93 | App.prototype.onHashChange = function onHashChange(e) { 94 | e.preventDefault(); 95 | var newHash = window.location.hash; 96 | console.log(e.oldURL); 97 | if (!e.oldURL) { 98 | console.info('HashChange (initial) -> ', newHash); 99 | } else { 100 | var oldHash = e.oldURL.substr(e.oldURL.indexOf('#')); 101 | console.info('HashChange', oldHash, ' -> ', newHash); 102 | if (oldHash === newHash) { 103 | return; 104 | } 105 | if (this.confirmExit()) { 106 | window.location.hash = '#' + this.currentPage.imageUrl; 107 | return; 108 | } 109 | } 110 | if (newHash.length > 2) 111 | this.loadPage(newHash.substr(1)); 112 | }; 113 | 114 | App.prototype.showHistory = function() { 115 | var self = this; 116 | this.history.load(function(err) { 117 | if (err) { 118 | return self.emit('app:ajaxError', err); 119 | } 120 | self.historyView.render(); 121 | }); 122 | }; 123 | 124 | App.prototype.render = function() { 125 | 126 | var self = this; 127 | 128 | // Select mode initially off 129 | this.selectMode = false; 130 | 131 | // Setup event handlers for drag and drop 132 | // TODO 133 | setupDragAndDrop(); 134 | 135 | // Render views 136 | this.waitingAnimation.render(); 137 | this.cheatsheetView.render(); 138 | this.toolbar.render(); 139 | this.selectbar.render(); 140 | this.dropzone.render(); 141 | 142 | this.on('app:loading', function hideSidebar() { self.sidebar.$el.addClass('hidden'); }); 143 | this.on('app:loading', function hidePageView() { self.pageView.$el.addClass('hidden'); }); 144 | this.on('app:loaded', function showSidebar() { self.sidebar.$el.removeClass('hidden'); }); 145 | this.on('app:loaded', function showPageView() { self.pageView.$el.removeClass('hidden'); }); 146 | }; 147 | 148 | App.prototype.init = function init() { 149 | var self = this; 150 | 151 | // window global events 152 | window.onbeforeunload = self.confirmExit.bind(self); 153 | window.onhashchange = self.onHashChange.bind(self); 154 | 155 | // Set up views 156 | this.pageView = new PageView({'el': "#file-correction",}); 157 | this.dropzone = new Dropzone({'el': '#dropzone'}); 158 | this.toolbar = new Toolbar({'el': '#toolbar'}); 159 | this.sidebar = new Sidebar({'el': '#right-sidebar'}); 160 | this.selectbar = new Selectbar({'el': '#select-bar'}); 161 | this.waitingAnimation = new WaitingAnimation({ 162 | 'el': "#waiting-animation", 163 | 'model': this.cheatsheet 164 | }); 165 | this.historyView = new HistoryView({ 166 | 'el': "#history-modal", 167 | 'model': this.history, 168 | 'tpl': this.templates.historyItem, 169 | }); 170 | this.cheatsheetView = new CheatsheetView({ 171 | 'el': "#cheatsheet-modal", 172 | 'model': this.cheatsheet, 173 | 'tpl': this.templates.cheatsheetEntry, 174 | }); 175 | 176 | // Load cheatsheet and errorTags 177 | async.each([this.cheatsheet, this.errorTags], function(model, done) { 178 | model.load(done); 179 | }, function(err) { 180 | if (err) return self.emit('app:ajaxError', err); 181 | self.settings.load(); 182 | self.render(); 183 | // Trigger hash change 184 | $(window).trigger('hashchange'); 185 | self.$el.trigger('app:initialized'); 186 | }); 187 | }; 188 | 189 | App.prototype.savePage = function savePage() { 190 | var self = this; 191 | let currentPage = window.app.currentPage; 192 | if (!currentPage) { 193 | notie.alert(1, "Nichts zu speichern", 1); 194 | } else { 195 | this.emit('app:saving'); 196 | window.app.currentPage.save(function(err) { 197 | if (err) { 198 | self.emit('app:ajaxError', err); 199 | } else { 200 | self.emit('app:saved'); 201 | } 202 | }); 203 | } 204 | }; 205 | 206 | App.prototype.loadPage = function loadPage(url) { 207 | var self = this; 208 | if (self.confirmExit()) return; 209 | this.emit('app:loading'); 210 | this.currentPage = new Page(url); 211 | this.currentPage.load(function(err) { 212 | if (err) { 213 | return self.emit('app:ajaxError', err); 214 | } 215 | self.pageView.model = self.currentPage; 216 | self.pageView.render(); 217 | self.sidebar.model = self.currentPage; 218 | self.sidebar.render(); 219 | self.emit('app:loaded'); 220 | }); 221 | }; 222 | -------------------------------------------------------------------------------- /doc/user-scripts/scrape-wiki.user.js: -------------------------------------------------------------------------------- 1 | // ==UserScript== 2 | // @name Extract Special Characters 3 | // @namespace http://github.com/kba/ 4 | // @include https://github.com/UB-Mannheim/ocr-gt-tools/wiki/Special-Characters 5 | // @include https://github.com/UB-Mannheim/ocr-gt-tools/wiki/Error-Tags 6 | // @description Extract special character data from ocr-gt-tools wiki 7 | // @version 1 8 | // @require https://code.jquery.com/jquery-2.2.3.min.js 9 | // @require https://cdnjs.cloudflare.com/ajax/libs/z-schema/3.17.0/ZSchema-browser.js 10 | // @grant GM_addStyle 11 | // @grant GM_setClipboard 12 | // ==/UserScript== 13 | /*globals GM_addStyle */ 14 | /*globals ZSchema */ 15 | 16 | var CSS = ` 17 | pre.schema-error 18 | { 19 | background: #a00; 20 | color: white; 21 | white-space: pre-wrap; 22 | } 23 | div#glyph-bar 24 | { 25 | font-size: x-large; 26 | position:fixed; 27 | bottom: 0; 28 | height: 48px; 29 | border: 2px solid black; 30 | background: white; 31 | width: 100%; 32 | } 33 | div#glyph-bar .left * { float: left; } 34 | div#glyph-bar .right * { float: right; } 35 | div#glyph-bar * 36 | { 37 | height: 100%; 38 | font-size: x-large; 39 | } 40 | div#glyph-bar input[type='text'] 41 | { 42 | font-family: "Garamond", "Bookman", serif; 43 | } 44 | div#schema-bar 45 | { 46 | position: fixed; 47 | z-index: 3000; 48 | top: 0; 49 | background: #900; 50 | color: white !important; 51 | width: 100%; 52 | font-size: x-large; 53 | height: 48px; 54 | border: 2px solid black; 55 | } 56 | div#schema-invalid 57 | { 58 | display: none; 59 | } 60 | div#schema-invalid a 61 | { 62 | display: inline-block; 63 | color: white !important; 64 | float: none; 65 | margin: 0 2px; 66 | } 67 | `; 68 | 69 | var SCHEMAS = { 70 | 'Special-Characters': { 71 | 'type': 'object', 72 | "additionalProperties": false, 73 | 'properties': { 74 | 'id': { 75 | 'type': 'string', 76 | 'pattern': '^[a-z0-9-]+$', 77 | }, 78 | 'sample': { 79 | 'type': 'array', 80 | 'items': { 81 | 'type': 'string', 82 | 'pattern': '^
164 |
165 |
166 |
167 |
168 | ${escapeHTML(JSON.stringify(err, null, 2))}`);
267 | $("#schema-invalid").show().append(
268 | `[${ $("#schema-invalid a").length + 1}]`);
269 | }
270 |
271 | $(function() {
272 | GM_addStyle(CSS);
273 | $("body").prepend(
274 | `
275 |
276 |
282 | `);
283 | var wikiPage = window.location.href.replace(/.*\//, '').replace(/#.*$/, '');
284 | var schema = SCHEMAS[wikiPage];
285 | var scraped = scrapeJsonFromWikiPage(schema);
286 | ON_LOAD[wikiPage](scraped);
287 | $("#copy-schema").on('click', function() {
288 | GM_setClipboard(JSON.stringify(SCHEMAS[schema], null, 2));
289 | window.alert("Copied JSON schema to clipboard");
290 | });
291 | $("#copy-json").on('click', function() {
292 | GM_setClipboard(JSON.stringify(scraped, null, 2));
293 | window.alert("Copied JSON schema to clipboard");
294 | });
295 | });
296 |
--------------------------------------------------------------------------------
/ocr-gt-tools.pug:
--------------------------------------------------------------------------------
1 | doctype html
2 | html(lang='de')
3 | head
4 | meta(charset='utf-8')
5 | title OCR-GT
6 | link(href='vendor.css', rel='stylesheet')
7 | link(rel='shortcut icon', type='image/x-icon', href='favicon.ico')
8 | link(href='ocr-gt-tools.css', rel='stylesheet')
9 | //- meta(http-equiv="Content-Type",content="text/html;charset=utf-8")
10 | body
11 |
12 | //*
13 | // Page info on the right
14 | //*
15 | script#tpl-rightSidebar(type='text/x-handlebars-template')
16 | .panel.panel-success
17 | .panel-heading
18 | h4.panel-title
19 | a(data-toggle='collapse', href='#page-comment') Seiten-Kommentar
20 | #page-comment.panel-collapse.collapse.in
21 | a.list-group-item(target='_blank', href='{{ this.url.hires-url }}')
22 | img(src='{{ this.url.thumb-url }}')
23 | textarea(placeholder="Seitenkommentar...") {{ page-comment }}
24 | .panel.panel-info
25 | .panel-heading
26 | h4.panel-title
27 | a(data-toggle='collapse', href='#work-info') Werk-Info
28 | #work-info.panel-collapse.collapse.in
29 | .list-group
30 | a.list-group-item(target='_blank', href='{{ this.url.landing-page-url }}')
31 | i.fa.fa-download
32 | | Werk in Digitaler Bibliothek
33 | | {{#each pages}}
34 | a.list-group-item(href='\#{{ this.url.thumb-url }}')
35 | i.fa.fa-history
36 | | Seite {{this.ids.page}}
37 | | {{/each}}
38 | .panel.panel-info
39 | .panel-heading
40 | h4.panel-title
41 | a(data-toggle='collapse', href='#page-info') Seiten-Info
42 | #page-info.panel-collapse.collapse.in
43 | .list-group
44 | a#download-comments.list-group-item(target='_blank', href='{{ this.url.comment-url }}')
45 | i.fa.fa-download
46 | | Anmerkungen
47 | a#download-correction.list-group-item(target='_blank', href='{{ this.url.correction-url }}')
48 | i.fa.fa-download
49 | | Korrektur
50 |
51 | //*
52 | // A cheatsheet entry
53 | //*
54 | script#tpl-cheatsheetEntry(type='text/x-handlebars-template')
55 | tr(id="cheatsheet-{{ id }}")
56 | td
57 | a(target="_blank", href="https://github.com/UB-Mannheim/ocr-gt-tools/wiki/Special-Characters\#{{ id }}")
58 | | {{ name.de }}
59 | td {{{ sample }}}
60 | td
61 | button.code(data-clipboard-text="{{ recognition }}")
62 | span.fa.fa-clipboard.clipboard
63 | span.recognition {{ recognition }}
64 |
65 | //*
66 | // A line of page
67 | //*
68 | script#tpl-lineContainer(type='text/x-handlebars-template')
69 | .row.line(id="line-{{ id }}")
70 |
71 | script#tpl-line(type='text/x-handlebars-template')
72 | .col-sm-1.button-col(data-target="#line-comment-{{ id }}")
73 | .btn-group.btn-group-sm(role='group')
74 | .btn-group(role='group')
75 | a.btn.btn-default.toggle-line-comment(data-target="#line-comment-{{ id }}")
76 | i.show-line-comment.fa.fa-commenting-o
77 | i.hide-line-comment.fa.fa-map-o.hidden
78 | .btn-group(role='group')
79 | button.btn.btn-default.dropdown-toggle(data-toggle="dropdown", type='button', aria-haspopup="true", aria-expanded="false", tabindex="-1")
80 | i.fa.fa-exclamation
81 | span.caret
82 | ul.dropdown-menu
83 | li
84 | a.btn.btn-default.add-comment(data-target="#line-comment-{{ id }}", data-tag="#text-italic")
85 | i.fa.fa-italic
86 | | Kursiv
87 | li
88 | a.btn.btn-warning.add-comment(data-target="#line-comment-{{ id }}", data-tag="#wrong-image-section")
89 | i.fa.fa-exclamation-circle
90 | | Fehler in Vorlage
91 |
92 | .col-sm-1.select-col.hidden(data-target="#line-comment-{{ id }}")
93 | input(type="checkbox")
94 |
95 | .col-sm-11.lines-col
96 | .panel.panel-default
97 | ul.list-group
98 | li.list-group-item.line-image
99 | img(src="{{ image }}", height="30")
100 | li.list-group-item.line-transcription
101 | input(type='text', value='{{ transcription }}')
102 | li.list-group-item.line-comment(id="line-comment-{{ id }}")
103 | textarea(placeholder="Kommentar zu dieser Zeile") {{ comment }}
104 |
105 | //*
106 | // A row of history
107 | //*
108 | script#tpl-historyItem(type='text/x-handlebars-template')
109 | tr
110 | td {{ date }}
111 | td
112 | a(href="\#{{ url }}") {{ url }}
113 | td {{ action }}
114 | td {{ ip }}
115 |
116 | //*
117 | // Modal popup containing the history
118 | //*
119 | #history-modal.modal.fade(tabindex="-1")
120 | .modal-admin
121 | .modal-content
122 | .modal-header
123 | h2.modal-title Verlauf
124 | button.close(type='button', data-dismiss='modal')
125 | span ×
126 | .modal-body
127 | table.table
128 | thead
129 | tr
130 | th Date
131 | th URL
132 | th Action
133 | th IP
134 | tbody
135 |
136 | //*
137 | // Help modal
138 | //*
139 | #cheatsheet-modal.modal.fade(tabindex="-1")
140 | .modal-admin
141 | .modal-content
142 | .modal-header
143 | h4.modal-title Eingabe Spickzettel
144 | .cheatsheet-entry
145 | input(type='text')
146 | button.close(type='button', data-dismiss='modal')
147 | span ×
148 | .modal-body
149 | table.table.table-condensed
150 | tbody.cheatsheet
151 |
152 | //*
153 | // Will contain the raw HTML
154 | //*
155 | #raw-html.hidden
156 |
157 | //*
158 | // Will contain the toolbar for the multi select mode
159 | //*
160 | #select-bar.hidden
161 | .panel-heading.clearfix
162 | button.toggle-select-mode.close.pull-left(type='button', data-dismiss='modal')
163 | span ×
164 | .btn-group.pull-right
165 | a.btn.btn-sm.btn-default.select-all
166 | i.fa.btn-sm.fa-arrows-alt
167 | | Alle auswählen
168 | a.btn.btn-sm.btn-default.select-none
169 | i.fa.btn-sm.fa-arrows-alt
170 | | Auswahl aufheben
171 | a.btn.btn-sm.btn-default.select-toggle
172 | i.fa.btn-sm.fa-arrows-alt
173 | | Auswahl umkehren
174 | a.btn.btn-sm.btn-default.add-multi-comment(data-tag="#text-italic")
175 | i.fa.btn-sm.fa-italic
176 | | Kursiv
177 | a.btn.btn-sm.btn-warning.add-multi-comment(data-tag="#wrong-image-section")
178 | i.fa.btn-sm.fa-exclamation-circle
179 | | Fehler in Vorlage
180 |
181 | //*
182 | // The navbar on the left side
183 | //*
184 | #toolbar.nav.navbar.navbar-default(role='navigation')
185 |
186 | // Brand and toggle get grouped for better mobile display
187 | .navbar-header
188 | button.navbar-toggle(type='button', data-toggle='collapse', data-target='.navbar-ex1-collapse')
189 | span.sr-only Toggle navigation
190 |
191 | // Collect the nav links, forms, and other content for toggling
192 | .collapse.navbar-collapse.navbar-ex1-collapse
193 | ul.nav.navbar-nav
194 | li
195 | button#save_button.btn.btn-success.disabled
196 | span#wait_save.hidden
197 | i.fa.fa-4x.fa-spinner.fa-pulse.fa-fw
198 | span#disk
199 | i.fa.fa-4x.fa-save.fa-fw
200 |
201 | //- Info
202 | li.dropdown.info
203 | a.btn.dropdown-toggle(data-toggle='dropdown')
204 | i.fa.fa-4x.fa-info-circle
205 | ul.dropdown-menu.navbar-default
206 | li
207 | .btn-group.btn-group-vertical
208 | a.btn.btn-default.pull-right(target="_blank", href="https://github.com/UB-Mannheim/ocr-gt-tools/wiki/Error-Tags")
209 | i.fa.fa-exclamation
210 | | Fehler-Kategorien
211 | button.btn.btn-default(data-toggle="modal",data-target="#cheatsheet-modal",data-keyboard="true")
212 | i.fa.fa-question
213 | | Spickzettel
214 | button.btn.btn-default(data-toggle="modal",data-target="#history-modal",data-keyboard="true")
215 | i.fa.fa-history
216 | | Verlauf
217 |
218 | //- Zoom
219 | li.dropdown.disabled.zoom
220 | a.btn.dropdown-toggle(data-toggle='dropdown')
221 | i.fa.fa-search-plus.fa-4x
222 | ul.dropdown-menu.navbar-default
223 | li
224 | .btn-group.btn-group-vertical
225 | button#zoom-in.btn.btn-default(type='button')
226 | i.fa.fa-search-plus
227 | | Größer
228 | button#zoom-reset.btn.btn-default(type='button')
229 | i.fa.fa-search
230 | | Zurücksetzen
231 | button#zoom-out.btn.btn-default(type='button')
232 | i.fa.fa-search-minus
233 | | Kleiner
234 |
235 | //- Show/hide comments
236 | li.dropdown.disabled.views
237 | a.btn.dropdown-toggle(data-toggle='dropdown')
238 | i.fa.fa-4x.fa-eye
239 | ul.dropdown-menu.navbar-default
240 | li
241 | .btn-group.btn-group-vertical
242 | button.set-view.btn.btn-default(type='button', data-target='*')
243 | i.fa.fa-picture-o
244 | i.fa.fa-commenting-o
245 | i.fa.fa-align-left
246 | | Vollansicht
247 | button.set-view.btn.btn-default(type='button', data-target='.line-transcription,.line-image')
248 | i.fa.fa-picture-o
249 | i.fa.fa-align-left
250 | | Bild + Text
251 | button.set-view.btn.btn-default(type='button', data-target='.line-comment,.line-image')
252 | i.fa.fa-picture-o
253 | i.fa.fa-commenting-o
254 | | Bild + Kommentar
255 | button.set-view.btn.btn-default(type='button', data-target='.line-comment,.line-transcription')
256 | i.fa.fa-align-left
257 | i.fa.fa-commenting-o
258 | | Text + Kommentar
259 | button.set-view.btn.btn-default(type='button', data-target='.line-image')
260 | i.fa.fa-picture-o
261 | | Bild
262 | button.set-view.btn.btn-default(type='button', data-target='.line-transcription')
263 | i.fa.fa-align-left
264 | | Text
265 | button.set-view.btn.btn-default(type='button', data-target='.line-comment')
266 | i.fa.fa-commenting-o
267 | | Kommentar
268 |
269 | li.dropdown.disabled.filter
270 | a.toggle-select-mode.btn.dropdown-toggle(data-toggle='dropdown')
271 | i.fa.fa-4x.fa-plus-square-o
272 |
273 | //- Filter
274 | li.dropdown.disabled.filter
275 | a.btn.dropdown-toggle(data-toggle='dropdown')
276 | i.fa.fa-4x.fa-sort
277 | ul.dropdown-menu.navbar-default
278 | li
279 | .btn-group.btn-group-vertical
280 | button#sort-line.btn.btn-default
281 | i.fa.fa-sort
282 | | Normale Sortierung (auf)
283 | button#sort-line-desc.btn.btn-default
284 | i.fa.fa-sort
285 | | Umgekehrte Sortierung (ab)
286 | button#sort-width.btn.btn-default
287 | i.fa.fa-sort
288 | | Nach Breite sortieren (auf)
289 | button#sort-width-desc.btn.btn-default
290 | i.fa.fa-sort
291 | | Nach Breite sortieren (ab)
292 |
293 | //*
294 | //- Here be dragons
295 | //*
296 | .container-fluid
297 | .row
298 | .col-sm-9#root
299 | #dropzone.jumbotron
300 | span#wait-load.hidden
301 | i.fa.fa-spinner.fa-pulse.fa-fw
302 | | Grafik auf dem Dokument ablegen
303 | br
304 | | oder
305 | br
306 | | URL angeben:
307 | br
308 | div.input-group#load-image
309 | input.input-lg.form-control(type='text')
310 | span.input-group-btn
311 | button.btn.btn-default.btn-lg(type="button") Los!
312 | #file-correction
313 | #waiting-animation.hidden
314 | #right-sidebar.col-sm-3.panel-group
315 |
316 | script(src='vendor.js')
317 | script(src='ocr-gt-tools.js')
318 | //- vim: sw=2 ts=2 noet ft=pug :
319 |
--------------------------------------------------------------------------------
/ocr-gt-tools.cgi:
--------------------------------------------------------------------------------
1 | #!/usr/bin/perl
2 | use strict;
3 | use warnings;
4 | use CGI;
5 | use CGI::Carp qw(carpout);
6 | use Cwd qw(abs_path);
7 | use Hash::Merge;
8 | use File::Basename qw(dirname);
9 | use File::Path qw(make_path);
10 | use IPC::Run qw(run);
11 | use JSON;
12 | use POSIX qw(strftime);
13 | use Time::HiRes qw(time);
14 | use YAML::XS qw(LoadFile Dump);
15 | use open ":encoding(utf8)";
16 |
17 | my $DEBUGLOG;
18 | my $DATE_FORMAT = "%Y-%m-%d";
19 | my $TIME_FORMAT = "%H:%M:%S";
20 |
21 | # Directory containing the CGI script
22 | my $OCR_GT_BASEDIR = dirname(abs_path($0));
23 | my $cgi = CGI->new;
24 | my $config = loadConfig();
25 |
26 | =head2 setupLogging
27 |
28 | Setup logging
29 |
30 | =cut
31 |
32 | sub setupLogging
33 | {
34 | my $logdir = $config->{'logging'}->{'logdir'};
35 | if (! -d "$logdir") {
36 | make_path "$logdir", {mode => oct(777)};
37 | }
38 | if ($config->{'logging'}->{'stderr'}) {
39 | $DEBUGLOG = *STDERR;
40 | } else {
41 | open($DEBUGLOG, ">>", "$logdir/ocr-gt-tools.log")
42 | or die "Cannot write to log file '$logdir/ocr-gt-tools.log': $!\n";
43 | }
44 | carpout(*$DEBUGLOG);
45 | }
46 |
47 | =head2 debug
48 |
49 | Log a message to the log file.
50 |
51 | =cut
52 |
53 | sub debug
54 | {
55 | my $msg = sprintf(shift(), @_);
56 | my $t = time;
57 | my $timestamp = strftime $TIME_FORMAT, localtime $t;
58 | $timestamp .= sprintf ".%03d", ($t-int($t))*1000; # without rounding
59 | printf $DEBUGLOG "%s: %s\n", $timestamp, $msg;
60 | }
61 |
62 |
63 | =head2 slurp
64 |
65 | Read file contents into a string.
66 |
67 | =cut
68 |
69 | sub slurp
70 | {
71 | my ($fname) = @_;
72 | my $out = '';
73 | open my $fh, "<:encoding(UTF-8)", $fname or httpError(500, "Could not read file %s: $!", $fname);
74 | $out .= $_ while(<$fh>);
75 | close $fh;
76 | chomp $out;
77 | return $out;
78 | }
79 |
80 | =head2 logRequest
81 |
82 | Log the IP and scan URL to request.log
83 |
84 | =cut
85 |
86 | sub logRequest
87 | {
88 | my $url = $cgi->param('imageUrl');
89 | if (!$url) {
90 | debug("No URL to log for this request");
91 | return;
92 | }
93 | my $action = $cgi->url_param('action');
94 | my $t = time;
95 | my $timestamp = strftime sprintf("%sT%sZ", $DATE_FORMAT, $TIME_FORMAT), localtime $t;
96 | my $json = JSON->new->utf8->pretty(0)->encode({
97 | date => $timestamp,
98 | action => $action,
99 | url => $url,
100 | ip => $ENV{REMOTE_ADDR}
101 | });
102 | my $logdir = $config->{'logging'}->{'logdir'};
103 | open(my $REQUESTLOG, ">>", "$logdir/request.log")
104 | or httpError(500, "Cannot write to log file '$logdir/request.log': $!\n");
105 | print $REQUESTLOG $json . "\n";
106 | close $REQUESTLOG;
107 | }
108 |
109 |
110 | =head2 httpError
111 |
112 | Send an HTTP error message
113 |
114 | =cut
115 |
116 | sub httpError
117 | {
118 | my $status = shift;
119 | my $msg = sprintf(shift(), @_);
120 | print $cgi->header(-type => 'text/plain', -status => $status);
121 | debug("********* ERROR $status - $msg *********");
122 | print $msg;
123 | exit 1;
124 | }
125 |
126 |
127 | =head2 httpJSON
128 |
129 | convert data to JSON and send it
130 |
131 | =cut
132 |
133 | sub httpJSON
134 | {
135 | my ($obj, $compact) = @_;
136 | if (ref $obj eq 'ARRAY') {
137 | $obj = [
138 | map { delete $_->{'path'}; delete $_->{'command'}; $_; }
139 | @{ $obj }
140 | ];
141 | } elsif (ref $obj) {
142 | delete $obj->{'path'}; delete $obj->{'command'}
143 | }
144 | my $json = JSON->new->utf8->pretty(1)->encode($obj);
145 | print $cgi->header( -type => 'application/json', -charset => 'utf-8');
146 | print $json;
147 | }
148 |
149 |
150 | =head2 loadConfig
151 |
152 | Load the configuration from the YAML file
153 |
154 | =cut
155 |
156 | sub loadConfig
157 | {
158 | my $ymlFile = "$OCR_GT_BASEDIR/ocr-gt-tools.default.yml";
159 | my $config = LoadFile($ymlFile);
160 |
161 | # Merge with non-default config
162 | if (-e "$OCR_GT_BASEDIR/ocr-gt-tools.yml") {
163 | my $merge = Hash::Merge->new('RIGHT_PRECEDENT');
164 | $config = $merge->merge($config, LoadFile("$OCR_GT_BASEDIR/ocr-gt-tools.yml"));
165 | }
166 |
167 | # Compile all 'pattern' as regexes
168 | for (my $i = 0; $i < scalar @{$config->{'pattern'}}; $i++) {
169 | my $pat = $config->{'pattern'}->[$i];
170 | $config->{'pattern'}->[$i] = qr/$pat/smx;
171 | }
172 |
173 | # Remove all newlines from 'template'
174 | for my $category (keys %{$config->{'template'}}) {
175 | while (my ($k, $v) = each %{$config->{'template'}->{$category}}) {
176 | $v =~ s/\n//g;
177 | $config->{'template'}->{$category}->{$k} = $v;
178 | }
179 | }
180 |
181 | # Replace !! INVALID
277 |
278 |
279 |
280 |
281 |
103 |
126 |
127 |
128 |
129 |
104 |
124 |
125 |
130 |
147 |
148 |
149 |
150 |
151 |
152 |
153 |
154 |
131 |
145 |
146 |
155 |
160 |
161 |
162 |
163 |
226 |
227 |
228 |
156 |
157 |
158 |
159 |
229 |
243 |
244 |
245 |
246 |
--------------------------------------------------------------------------------
/dist/example/fileadmin/417576986/hocr/417576986_0013.hocr:
--------------------------------------------------------------------------------
1 |
2 |
4 |
5 |
6 |
230 |
242 |
231 |
240 |
241 | Grafik auf dem Dokument ablegen
oder
URL angeben:
232 |
237 |
238 |
239 | oder
URL angeben:
232 |
233 |
234 |
235 |
236 |
13 |
82 |
83 |
84 |
--------------------------------------------------------------------------------
14 |
75 | 15 | P R E F A C E. 16 | 17 | dillinguer les principes des 18 | 19 | confequences & les regles des 20 | 21 | exceprjons s & c’et’t ce qu: fait 22 | 23 | une infljrution Il y a long rems 24 | 25 | que j’en voi la necelllre & que 26 | 27 | je desire qu’jl y en ait en tou- 28 | 29 | tes les marieres qu’jl importe 30 | 31 | de favoir. Ceft aum ce qui m’a 32 | 33 | porte ä compol'er le Carechle 34 | 35 | me historjque & le trajtä de la 36 | 37 | merhode des etudes. Sans ce ('e- 38 | 39 | cours on marche ä tärons , on 40 | 41 | commence par de petits details, 42 | 43 | on fuit Paurorjte du premier 44 | 45 | venu , on ne forme que des 46 | 47 | doutes sc des opinions incertai- 48 | 49 | nes. 50 | 51 |
52 | 53 |54 | Tel eft l*etat des pures prati- 55 | 56 | ciens, qui n'apprennent la juer- 57 | 58 | pruclence canonjque que com- 59 | 60 | meles artjfans apprcnnent les 61 | 62 | metjers leg plus vle Sen voyant 63 | 64 | travailler leurs mairres , & re- 65 | 66 | tenant ce quïls leur difent ä 67 | 68 |
69 | 70 |71 | lll] 72 | 73 |
74 |
76 |
81 | 77 | 78 | 79 |
80 |