Preliminary support for corrected text

* It's slow, need to figure out how to load it in the background maybe
* The bouding boxes could be improved

3 files changed, 19 insertions, 9 deletions

diff --git a/web/main.py b/web/main.py
index 582d33c..206eb86 100644
--- a/web/main.py
+++ b/web/main.py
@@ -3,7 +3,7 @@ from tornado.web import RequestHandler, Application
 import tornado.ioloop
 from settings import settings
 import utils
-from parsedjvutext import image_from_book
+from djvu_utils import image_from_book
 import io
 
 class MainHandler(RequestHandler):
@@ -16,7 +16,7 @@ class MainHandler(RequestHandler):
 class ImageHandler(RequestHandler):
 
     def get(self, page_number):
-        im = image_from_book(self.settings["book"], int(page_number))
+        im = image_from_book("../" + self.settings["book"], int(page_number))
         self.set_header('Content-Type', 'image/jpg')
         img_buff = io.BytesIO()
         im.save(img_buff, format="JPEG")
diff --git a/web/settings.py b/web/settings.py
index 98c490b..5a8c9aa 100644
--- a/web/settings.py
+++ b/web/settings.py
@@ -4,5 +4,6 @@ settings = {
     "static_path": "static",
     "cookie_secret": "toto",
     "login_url": "/login",
-    "book": "../Bloy_-_Le_Sang_du_pauvre,_Stock,_1932.djvu"
+    #"book": "Bloy_-_Le_Sang_du_pauvre,_Stock,_1932.djvu"
+    "book": "Villiers_de_L'Isle-Adam_-_Tribulat_Bonhomet,_1908.djvu"
 }
diff --git a/web/utils.py b/web/utils.py
index bb9a4fe..8522841 100644
--- a/web/utils.py
+++ b/web/utils.py
@@ -1,12 +1,21 @@
-from parsedjvutext import parse_book
+import djvu_utils as du
 import sys
-
+import string_utils as su
+from wikisource import get_page
 
 def gen_html(book, page_number):
-    d = parse_book(book, page=int(page_number), html=True)
-    if d[0]:
-        words, coords = zip(*d[0])
-        return (list(enumerate(coords)), list(enumerate(words)))
+    doc = du.get_document("../" + book)
+    page = doc.pages[int(page_number)-1]
+    d = du.parse_page(page)
+    corrected_text = get_page(book, int(page_number))
+    corrected_words = su.simplify(corrected_text).split()
+    if d:
+        words, coords = zip(*d)
+        C = su.align(corrected_words, list(words), list(coords))
+        r = su.alignment_to_sexp(corrected_text.split(), words, coords, C[1])
+        corrected_words, coords = zip(*r)
+        coords_html = du.convert_to_htmlcoord(coords, page.size[1])
+    return (list(enumerate(coords_html)), list(enumerate(corrected_words)))
 
 if __name__ == "__main__":
     gen_html(*sys.argv[1:3])