Merge pull request #1 from x4base/master

使用tesseract做初步的OCR

toufu/README.md

@@ -33,4 +33,13 @@
     done
 
 
+    使用tesseract 做中英文的初步OCR，可以讓大家少打一點字
+
+    需要先在系統上安裝好tesseract, 並裝好中文語言檔(chi_tra)，另外db裡toufu這個表要有ocr_eng, ocr_cht這兩欄
+
+	    for n in *.jpg; do
+	      php ocr_guess.php "$n";
+	    done
+
+
 4. 寫前端，從 https://github.com/ctiml/campaign-finance.g0v.ctiml.tw 抄一些設計過來

toufu/cell.js

@@ -55,6 +55,9 @@ $(document).ready(function(){
         .text("")
         .append($('<span></span>').text("第 "+res.p+" 頁, 第 "+res.line+" 行"));
 
+          $('#ocrEng').text(res.ocr_eng);
+          $('#ocrCht').text(res.ocr_cht);
+
         if (res.ans !== null) {
           $('.cell-info').append($('<span></span>').text(" 已經有" +res.cnt + "人填寫確認了。"));
           $('.confirm').show();
@@ -72,6 +75,8 @@ $(document).ready(function(){
     $('.confirm').hide();
     $('.cell-image').html("");
     $('#unclear').hide();
+    $('#ocrEng').text('');
+    $('#ocrCht').text('');
 
     if (question_pools.length) {
         set_question(question_pools.shift());

toufu/index.php

@@ -44,6 +44,16 @@
     <div class="bar" style="width: 80%;"></div>
     <span id="progress_text"></span>
     </div>
+    <div><b>以下是供參考的機器自動辨識結果，可以複製正確的部份到答案欄，可以少打一些字</b></div>
+    <div class="ocr_block">
+        <label for="">英文OCR</label>
+        <textarea name="" id="ocrEng" cols="50" rows="4"></textarea>
+    </div>
+    <div class="ocr_block">
+        <label for="">中文OCR</label>
+        <textarea name="" id="ocrCht" cols="50" rows="4"></textarea>
+    </div>
+    <div style="clear: both;"></div>
 </div>
 <script src="cell.js"></script>
 <link rel="stylesheet" href="cell.css">

toufu/ocr_guess.php

@@ -0,0 +1,43 @@
+<?php
+
+if (!isset($argv[1])) {
+    echo 'please supply image file name';
+    exit(1);
+}
+
+$imgFn = $argv[1];
+$tmpFileBase = '/tmp/amisOcrGuess';
+$tmpFile = $tmpFileBase . ".txt";
+
+function getOcr($lang)
+{
+    global $imgFn, $tmpFileBase, $tmpFile;
+
+    $cmd = 'tesseract "' . $imgFn . '"  "' . $tmpFileBase . '" -l ' . $lang . ' 2>&1 > /dev/null';
+    exec($cmd);
+    $ocr = file_get_contents($tmpFile);
+    $ocr = trim($ocr);
+
+    return $ocr;
+}
+
+preg_match('/([0-9]{3})_([0-9]{3}).jpg/', $imgFn, $matches);
+if (!$matches) {
+    die('wrong format of the image file name ');
+}
+$p = $matches[1];
+$line = $matches[2];
+
+$ocrEng = getOcr('eng');
+$ocrCht = getOcr('chi_tra');
+
+$pdo = new PDO("sqlite:toufu.sq3");
+$st = $pdo->prepare("UPDATE toufu SET ocr_eng=:eng, ocr_cht=:cht WHERE p=:p AND line=:line ");
+$st->execute(array(
+    'eng' => $ocrEng,
+    'cht' => $ocrCht,
+    'p' => $p,
+    'line' => $line
+));
+
+echo "$p $line \n";

toufu/style.css

@@ -1,3 +1,13 @@
 body {
     padding-top: 20px;
 }
+
+.ocr_block{
+    float: left;
+    margin: 10px
+}
+
+.ocr_block textarea{
+    width: 400px;
+    height: 130px;
+}