fix: remove dead next/prev links when merging utterances

src/merge_utterances.py

@@ -3,9 +3,13 @@
 """
 from lxml import etree
 import argparse
-from pyriksdagen.utils import protocol_iterators
+from pyriksdagen.utils import corpus_iterator
 from tqdm import tqdm
 from pyriksdagen.utils import TEI_NS
+from pyriksdagen.io import (
+    parse_tei,
+    write_tei
+)
 
 def merge_us(root):
     for body in root.findall(f".//{TEI_NS}body"):
@@ -23,18 +27,25 @@ def merge_us(root):
                     previous_u = None
     return root
 
+def remove_dead_nextprev_links(root):
+    all_ids = set(root.xpath('//@xml:id', namespaces={"xml": "http://www.w3.org/XML/1998/namespace"}))
+    for body in root.findall(f".//{TEI_NS}body"):
+        for div in body.findall(f".//{TEI_NS}div"):
+            for elem in div:
+                if elem.tag.split("}")[-1] == "u":
+                    if elem.attrib.get("next") not in all_ids and "next" in elem.attrib:
+                        del elem.attrib["next"]
+                    if elem.attrib.get("prev") not in all_ids and "prev" in elem.attrib:
+                        del elem.attrib["prev"]
+    return root
+
 def main(args):
-    protocols = sorted(list(protocol_iterators(args.records_folder, start=args.start, end=args.end)))
-    parser = etree.XMLParser(remove_blank_text=True)
+    protocols = sorted(list(corpus_iterator("prot", args.records_folder, start=args.start, end=args.end)))
     for p in tqdm(protocols):
-        root = etree.parse(p, parser).getroot()
+        root, ns = parse_tei(p)
         root = merge_us(root)
-
-        b = etree.tostring(
-            root, pretty_print=True, encoding="utf-8", xml_declaration=True
-        )
-        with open(p, "wb") as f:
-            f.write(b)
+        root = remove_dead_nextprev_links(root)
+        write_tei(root, p)
 
 
 if __name__ == "__main__":