Merge branch 'master' of horel.org:thibaut/ocr-layer-curation

Conflicts:
	web/static/css/style.css
	web/templates/index.html
	web/utils.py

6 files changed, 80 insertions, 38 deletions

diff --git a/string_utils.py b/string_utils.py
index a8a38c0..0588418 100644
--- a/string_utils.py
+++ b/string_utils.py
@@ -156,18 +156,31 @@ def print_alignment(l1, l2, c2, alignment):
         for word in l2:
             print u"{0:>25} | {1}".format("", word)
 
-def alignment_to_sexp(l1, l2, c2, alignment):
+def invert_align(alignment, n):
+    l = [[] for _ in range(n)]
+    for i, e in enumerate(alignment):
+        for a in e:
+            l[a].append(i)
+    return l
+
+def alignment_to_coord(l1, alignment):
+    # l1 list of corrected words
+    # alignment list of size len(l1) qui mappe mots dans l2
+    # returns indices in l2
+
     r = []
     prev = 0
     for index, g in itertools.groupby(zip(l1, alignment), lambda x:x[1]):
         word = " ".join([a[0] for a in g])
-        if not index:
-            r.append([word, []])
-        else:
-            begin, end = index[0], index[-1]
-            if end > begin:
-                #need to find a better way to get the box coordinates
-                r.append([word, c2[begin]])
-            else:
-                r.append([word, c2[begin]])
+        r.append([word, index])
+        # if not index:
+        #     r.append([word, None])
+        # else:
+
+        #     begin, end = index[0], index[-1]
+        #     if end > begin:
+        #         #need to find a better way to get the box coordinates
+        #         r.append([word, begin])
+        #     else:
+        #         r.append([word, begin])
     return r
diff --git a/web/main.py b/web/main.py
index 206eb86..a6826c1 100644
--- a/web/main.py
+++ b/web/main.py
@@ -9,9 +9,10 @@ import io
 class MainHandler(RequestHandler):
 
     def get(self, page_number):
-        areas, words = utils.gen_html(self.settings["book"], page_number)
-        self.render("index.html", page_number=page_number,
-                    areas=areas, words=words)
+        orig_coords, orig_words, corr_words, align = \
+            utils.gen_html(self.settings["book"], page_number)
+        self.render("index.html", page_number=page_number, orig_coords=orig_coords,
+                    orig_words=orig_words, corr_words=corr_words, align=align)
 
 class ImageHandler(RequestHandler):
 
diff --git a/web/static/css/style.css b/web/static/css/style.css
index ec73fd0..b293da8 100644
--- a/web/static/css/style.css
+++ b/web/static/css/style.css
@@ -27,4 +27,15 @@ span:hover {
     padding: 3.5em 2em;
     font-size: 18px;
     line-height: 180%;
+
+#texte-non-corrige {
+    margin-top:0cm;
+    width: 10cm;
+    float: left;
+}
+
+#texte-corrige {
+    margin-top:0cm;
+    width: 10cm;
+    float: left;
 }
diff --git a/web/static/js/main.js b/web/static/js/main.js
index ff09fd1..c790e34 100644
--- a/web/static/js/main.js
+++ b/web/static/js/main.js
@@ -1,19 +1,23 @@
 $(document).ready(function() {
     $('#page').mapster({
-                mapKey: 'data-id',
-                fillColor: 'ff0000',
-                fillOpacity: 0.3, 
-    onMouseover: function (e) {
-        $("#" + e.key).addClass("selected");
-    },
-    onMouseout: function (e) {
-        $("#" + e.key).removeClass("selected");
-    }
-                
+        mapKey: 'data-orig',
+        fillColor: 'ff0000',
+        fillOpacity: 0.3,
+        onMouseover: function (e) {
+            $("#" + "orig-" + e.key).addClass("selected");
+            $("#" + "corr-" + e.key).addClass("selected");
+        },
+        onMouseout: function (e) {
+            $("#" + "orig-" + e.key).removeClass("selected");
+            $("#" + "corr-" + e.key).removeClass("selected");
+        }
+
     }).mapster('resize', 500);
 
     $("span").mouseenter(function() {
-        $('area[data-id='+$(this).attr("id")+']').mapster("highlight");
+        $(this).attr("id").replace(/\D+/,"").split(",").map(function(e){
+            $('area[data-orig='+e+']').mapster("highlight");
+        })
     });
 
     $("span").mouseout(function() {
diff --git a/web/templates/index.html b/web/templates/index.html
index dc286aa..0a07be1 100644
--- a/web/templates/index.html
+++ b/web/templates/index.html
@@ -1,13 +1,26 @@
 {% extends "layout.html" %}
 
 {% block main %}
-    <div id="image_container">
-        <img id="page" src="{{page_number}}.jpg" usemap="#wordmap" />
-    </div>
-    <map name="wordmap">{% for id, coords in areas %}
-        <area href="#" shape="rect" coords="{{coords}}" data-id="word-{{id}}" />{% end %}
-    </map>
-    <div id="text">
-        {% raw words %}
-    </div>
+<div id="image_container">
+  <img id="page" src="{{page_number}}.jpg" usemap="#wordmap" />
+</div>
+<map name="wordmap">
+  {% for i, coord in enumerate(orig_coords) %}
+  <area href="#" shape="rect" coords="{{coord}}" data-orig="{{i}}" />
+        {# data-corr="{{",".join(map(str,corr_coords_index[i]))}}" #}
+  {% end %}
+</map>
+
+<div id="texte-non-corrige">
+  <h3>Texte original</h3>
+  {% for i, word in enumerate(orig_words) %}
+  <span id="orig-{{i}}">{{word}}</span>
+  {% end %}
+</div>
+<div id="texte-corrige">
+  <h3>Texte corrigé</h3>
+  {% for i, word in enumerate(corr_words) %}
+  <span id="corr-{{",".join(map(str,align[i]))}}">{{word}}</span>
+  {% end %}
+</div>
 {% end %}
diff --git a/web/utils.py b/web/utils.py
index 1947f8b..e6f4309 100644
--- a/web/utils.py
+++ b/web/utils.py
@@ -10,11 +10,11 @@ def gen_html(book, page_number):
     d = du.parse_page(page)
     elem, corrected_text = get_page2(open("test.txt").read())
     if d:
-        words, coords = zip(*d)
-        C = su.align(corrected_text.split(), list(words), list(coords))
-        coords = [coords[e[0]] for e in C[1]]
-        coords_html = du.convert_to_htmlcoord(coords, page.size[1])
-    return (list(enumerate(coords_html)), str(elem))
+        orig_words, orig_coords = zip(*d)
+        C = su.align(corrected_words, list(orig_words), list(orig_coords))
+        corr_words = corrected_text.split()
+        orig_coords_html = du.convert_to_htmlcoord(orig_coords, page.size[1])
+    return orig_coords_html, orig_words, corr_words, C[1]
 
 if __name__ == "__main__":
     gen_html(*sys.argv[1:3])