Add inflections

add_inflections.py

@@ -0,0 +1,46 @@
+import sys
+from bs4 import BeautifulSoup
+
+# from https://github.com/coljac/kindle_eo_eng/blob/master/src/inflect.py
+def inflect(key):
+    # ekz: blankkola maraglo
+    if ' ' in key:
+        return []
+
+    root = key[:-1]
+    if key.endswith("o"):
+        return [root + x for x in ["on", "oj", "ojn"]]
+    elif key.endswith("a"):
+        return [root + x for x in ["aj", "an", "ajn"]]
+    elif key.endswith("e"):
+        return [root + "en"]
+    elif key.endswith("i"):
+        return [root + x for x in ["as", "os", "is", "us", "u", 
+            "ita", "ata", "ota", 
+            "inta", "anta", "onta", 
+            "intan", "antan", "ontan", 
+            "intaj", "antaj", "ontaj", 
+            "intajn", "antajn", "ontajn"]]
+    else:
+        return []
+
+def main(path: str, output_path: str):
+    with open(path) as f:
+        soup = BeautifulSoup(f, 'html.parser')
+
+    entries = soup.find_all('idx:orth')
+    for tag in entries:
+        tag['value'] = tag['value'].strip().strip(',')
+        inflections = inflect(tag['value'])
+        if inflections:
+            inflection_tag = soup.new_tag("idx:infl")
+            for inflection in inflections:
+                new_tag = soup.new_tag("idx:iform", attrs={'value': inflection})
+                inflection_tag.append(new_tag)
+            tag.append(inflection_tag)
+
+    with open(output_path, 'w') as f:
+        f.write(str(soup))
+
+if __name__ == '__main__':
+    main(sys.argv[1], sys.argv[2])

generate_opf.py

@@ -5,7 +5,7 @@ def main():
         lines = f.readlines()
 
     html_dir = Path('./output')
-    html_files = list(html_dir.glob("content*.html"))
+    html_files = list(html_dir.glob("inflected_content*.html"))
     html_files.sort()
 
     n = 0
@@ -14,8 +14,7 @@ def main():
         for line in lines:
             if "<item id" in line:
                 for n, html in enumerate(html_files):
-                    # TODO test media-type="application/xhtml+xml"
-                    f.write(f"""    <item id="dictionary{n}" href="{html.name}" media-type="text/x-oeb1-document"/>\n""")
+                    f.write(f"""    <item id="dictionary{n}" href="{html.name}" media-type="application/xhtml+xml"/>\n""")
             elif "<itemref" in line:
                 # the previous loop will set "n"
                 for n in range(n + 1):

process.sh

@@ -1,4 +1,9 @@
 #!/usr/bin/env bash
 xsltproc --path voko-grundo/dtd:revo-fonto/cfg xsl/revohtml.xsl revo-fonto/revo/*.xml > output/vortaro.html
 python split_html.py
+for HTML in output/content*.html
+do
+    HTML_INFLECTED=$(echo $HTML | sed -e 's/content/inflected_content/')
+    python add_inflections.py $HTML $HTML_INFLECTED
+done
 python generate_opf.py