From b31ace79ea5f6b3f279c544cd3a443d6fbf2a24d Mon Sep 17 00:00:00 2001
From: Patrick Simianer <patrick@lilt.com>
Date: Thu, 26 Feb 2026 10:05:59 +0000
Subject: overhaul

---
 tmx-extract | 75 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 75 insertions(+)
 create mode 100755 tmx-extract

(limited to 'tmx-extract')

diff --git a/tmx-extract b/tmx-extract
new file mode 100755
index 0000000..7791eb6
--- /dev/null
+++ b/tmx-extract
@@ -0,0 +1,75 @@
+#!/usr/bin/env python3
+#
+# Adapted from Apertium
+# http://wiki.apertium.org/wiki/Tools_for_TMX
+#
+
+from xml.sax import make_parser
+from xml.sax.handler import ContentHandler
+
+import sys
+import codecs
+
+class TMXHandler(ContentHandler):
+    def __init__ (self, slang, tlang, sfile, tfile):
+        self.pair = set([slang, tlang])
+        self.files = {}
+        self.files[slang] = sfile
+        self.files[tlang] = tfile
+        self.inTag = ""
+        self.note = ""
+        self.tuid = ""
+        self.type = ""
+        self.cur_pair = set()
+        self.cur_lang = ""
+        self.seg = {}
+        self.seg[slang] = ""
+        self.seg[tlang] = ""
+
+    def startElement(self, name, attrs):
+
+        if name == "tu":
+            self.cur_pair = set()
+            self.inTag = "tu"
+            self.tuid = attrs.get("tuid", "")
+            self.type = attrs.get("datatype", "")
+        elif name == "note":
+            self.inTag = "note"
+            self.note = ""
+        elif name == "tuv":
+            self.inTag = "tuv"
+            self.cur_lang = attrs.get("xml:lang", "")
+            self.cur_pair.add(self.cur_lang)
+        elif name == "seg":
+            self.inTag = "seg"
+            if self.cur_lang in self.pair:
+                self.seg[self.cur_lang] = ""
+
+    def characters (self, c):
+        if self.inTag == "note":
+            self.note += c
+        elif self.inTag == "seg" and self.cur_lang in self.pair:
+            self.seg[self.cur_lang] += c
+
+    def endElement(self, name):
+        if name == "tu" and self.pair == self.cur_pair:
+            for lang in self.cur_pair:
+                self.files[lang].write("{}\n".format(self.seg[lang].replace("\n", " ").strip()))
+
+
+if __name__ == "__main__":
+    parser = make_parser()
+
+    if len(sys.argv) < 3:
+        print(f"Usage: {sys.argv[0]} <file> <slang> <tlang>")
+        print()
+        sys.exit(-1)
+
+    sfile_path = f"{sys.argv[1]}.{sys.argv[2]}"
+    tfile_path = f"{sys.argv[1]}.{sys.argv[3]}"
+
+    with open(sfile_path, "w+") as sfile, open(tfile_path, "w+") as tfile:
+        curHandler = TMXHandler(sys.argv[2], sys.argv[3], sfile, tfile)
+        parser.setContentHandler(curHandler)
+        with open(sys.argv[1], "r") as tmx_file:
+            parser.parse(tmx_file)
-- 
cgit v1.2.3