convert languages, PRImA-Research-Lab/PAGE-XML#27, #3

OCR-D · Apr 8, 2021 · f1b67bd · f1b67bd
1 parent 570a917
commit f1b67bd
Show file tree

Hide file tree

Showing 7 changed files with 3,803 additions and 3 deletions.
diff --git a/ocrd_page_to_alto/convert.py b/ocrd_page_to_alto/convert.py
@@ -6,8 +6,9 @@
 
 from .utils import (
     set_alto_id_from_page_id,
-    set_alto_xywh_from_coords,
+    set_alto_lang_from_page_lang,
     set_alto_shape_from_coords,
+    set_alto_xywh_from_coords,
     setxml
 )
 from .styles import TextStylesManager
@@ -171,6 +172,7 @@ def _convert_textlines(self, reg_alto, reg_page):
             set_alto_id_from_page_id(line_alto, line_page)
             set_alto_xywh_from_coords(line_alto, line_page)
             set_alto_shape_from_coords(line_alto, line_page)
+            set_alto_lang_from_page_lang(line_alto, line_page)
             self.textstyle_mgr.set_alto_styleref_from_textstyle(line_alto, line_page)
             # XXX ALTO does not allow TextLine without at least one String
             if is_empty_line:
@@ -181,6 +183,7 @@ def _convert_textlines(self, reg_alto, reg_page):
                 set_alto_id_from_page_id(word_alto, word_page)
                 set_alto_xywh_from_coords(word_alto, word_page)
                 set_alto_shape_from_coords(word_alto, word_page)
+                set_alto_lang_from_page_lang(word_alto, word_page)
                 word_alto.set('CONTENT', word_page.get_TextEquiv()[0].get_Unicode())
 
     def _convert_table(self, parent_alto, parent_page, level=0):
@@ -191,11 +194,13 @@ def _convert_table(self, parent_alto, parent_page, level=0):
             if parent_page.get_TextRegion():
                 reg_alto = ET.SubElement(parent_alto, 'ComposedBlock')
                 set_alto_id_from_page_id(reg_alto, parent_page) # TODO not unique!
+                set_alto_lang_from_page_lang(reg_alto, parent_page)
                 for reg_page in parent_page.get_TextRegion():
                     self._convert_table(reg_alto, reg_page, level=level + 1)
             else:
                 textblock_alto = ET.SubElement(parent_alto, 'TextBlock')
                 set_alto_id_from_page_id(textblock_alto, parent_page)
+                set_alto_lang_from_page_lang(textblock_alto, parent_page)
                 self._convert_textlines(textblock_alto, parent_page)
 
     def convert_text(self):
@@ -208,6 +213,7 @@ def convert_text(self):
             set_alto_id_from_page_id(reg_alto, reg_page)
             set_alto_xywh_from_coords(reg_alto, reg_page)
             set_alto_shape_from_coords(reg_alto, reg_page)
+            set_alto_lang_from_page_lang(reg_alto, reg_page)
             self.textstyle_mgr.set_alto_styleref_from_textstyle(reg_alto, reg_page)
             if reg_page_type == 'Text':
                 self._convert_textlines(reg_alto, reg_page)

diff --git a/ocrd_page_to_alto/utils.py b/ocrd_page_to_alto/utils.py
@@ -1,5 +1,6 @@
 from lxml import etree as ET
 from ocrd_utils import xywh_from_points
+import langcodes
 
 def setxml(el, name, val):
     el.set(name, str(val))
@@ -20,3 +21,12 @@ def set_alto_shape_from_coords(reg_alto, reg_page):
 
 def set_alto_id_from_page_id(reg_alto, reg_page):
     setxml(reg_alto, 'ID', reg_page.id)
+
+def set_alto_lang_from_page_lang(reg_alto, reg_page):
+    for prefix in ('primaryL', 'secondaryL', 'l'):
+        lang_page = getattr(reg_page, f'{prefix}anguage', None)
+        if lang_page:
+            lang_alto = langcodes.find(lang_page).to_alpha3()
+            setxml(reg_alto, 'LANG',lang_alto)
+            return
+
diff --git a/requirements.txt b/requirements.txt
@@ -1,2 +1,3 @@
 ocrd >= 2.23.2
 lxml
+langcodes[data] >= 3.1.0
diff --git a/tests/data/alto-4-2.xsd b/tests/data/alto-4-2.xsd
diff --git a/tests/data/language.page.xml b/tests/data/language.page.xml
@@ -0,0 +1,23 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<pc:PcGts xmlns:pc="http://schema.primaresearch.org/PAGE/gts/pagecontent/2019-07-15" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://schema.primaresearch.org/PAGE/gts/pagecontent/2019-07-15 http://schema.primaresearch.org/PAGE/gts/pagecontent/2019-07-15/pagecontent.xsd" pcGtsId="OCR-D-OCR-CALAMARI_00001">
+    <pc:Page imageFilename="OCR-D-IMG/044417.jpg" imageWidth="3195" imageHeight="4370" type="content" primaryLanguage="Welsh" secondaryLanguage="Urdu" primaryScript="Avst - Avestan" secondaryScript="Cans - Unified Canadian Aboriginal Syllabics">
+        <pc:Border>
+            <pc:Coords points="61,254 2770,254 2770,4360 61,4360"/>
+        </pc:Border>
+        <pc:TextRegion id="r1" primaryLanguage="Volapük" secondaryLanguage="Interlingua" primaryScript="Cham - Cham" secondaryScript="Buhd - Buhid">
+            <pc:Coords points="0,0 1,1 1,0 0,1"/>
+            <pc:TextLine id="r1-l1" primaryLanguage="Norwegian Bokmål">
+                <pc:Coords points="0,0 1,1 1,0 0,1"/>
+                <pc:Word id="r1-l1-w1" language="Esperanto">
+                    <pc:Coords points="0,0 1,1 1,0 0,1"/>
+                    <pc:TextEquiv>
+                        <pc:Unicode>patrofikulo</pc:Unicode>
+                    </pc:TextEquiv>
+                </pc:Word>
+                <pc:TextEquiv>
+                    <pc:Unicode>patrofikulo</pc:Unicode>
+                </pc:TextEquiv>
+            </pc:TextLine>
+        </pc:TextRegion>
+    </pc:Page>
+</pc:PcGts>