Autofix: Add \p. Fixes Bridgeconn#270

kavitharaju · Nov 25, 2024 · f8a3f22 · f8a3f22
1 parent fd59749
commit f8a3f22
Show file tree

Hide file tree

Showing 4 changed files with 54 additions and 0 deletions.
diff --git a/node-usfm-parser/src/validator.js b/node-usfm-parser/src/validator.js
@@ -167,6 +167,15 @@ class Validator {
                 modifiedUSFM = modifiedUSFM.replace(toReplace, `${toReplace}\\p\n`);
                 changed = true;
             }
+            else if (error.isError && !errorText.startsWith("\\") && error.previousSibling.type === "chapter" &&
+                !error.children.some(ch => ch.type === "paragraph")) {
+                // console.log("Match 7.1");
+                const start = error.previousSibling.startIndex;
+                const end = error.startIndex;
+                const toReplace = modifiedUSFM.slice(start, end);
+                modifiedUSFM = modifiedUSFM.replace(toReplace, `${toReplace}\\p\n`);
+                changed = true;
+            }
             // Stray slash not with a valid marker
             else if (errorText.startsWith("\\") && !validMarkersPattern.test(errorText)) {
                 // console.log("Match 8");

diff --git a/py-usfm-parser/src/usfm_grammar/validator.py b/py-usfm-parser/src/usfm_grammar/validator.py
@@ -158,6 +158,18 @@ def auto_fix_usfm(self, usfm, fixed=False): #pylint: disable=R0912,R0915
                 # self.modified_usfm = re.sub(to_replace, repalcement, self.modified_usfm)
                 self.modified_usfm = self.modified_usfm.replace(to_replace, repalcement)
                 changed = True
+            elif error.is_error and \
+               (not error_text.startswith("\\")) and \
+               error.prev_sibling.type == "chapter" and \
+               "paragraph" not in [ch.type for ch in error.children]:
+                # print("match 7.1")
+                start = error.prev_sibling.start_byte
+                end = error.start_byte
+                to_replace = self.usfm_bytes[start:end].decode('utf-8')
+                repalcement = to_replace+"\\p\n"
+                # self.modified_usfm = re.sub(to_replace, repalcement, self.modified_usfm)
+                self.modified_usfm = self.modified_usfm.replace(to_replace, repalcement)
+                changed = True
             # Stray slash not with a valid marker
             elif error_text.startswith("\\") and \
                  (not re.match(valid_markers_pattern, error_text)):

diff --git a/tests/autofix/b_without_p.usfm b/tests/autofix/b_without_p.usfm
@@ -0,0 +1,24 @@
+\id GEN genesis Some desc
+\c 1
+\p
+\v 1 test verse
+\s some poem
+\q1
+first line
+\b
+\q1
+\v 2 more lines
+\b
+remaining verses
+\v 3 text follows
+\c 2
+\p
+\v 1 test verse
+\s some poem
+\q1
+first line
+\b
+\q1
+\v 2 more lines
+\b
+\v 3 verse follows
diff --git a/web-usfm-parser/src/validator.js b/web-usfm-parser/src/validator.js
@@ -178,6 +178,15 @@ class Validator {
                 modifiedUSFM = modifiedUSFM.replace(toReplace, `${toReplace}\\p\n`);
                 changed = true;
             }
+            else if (error.isError && !errorText.startsWith("\\") && error.previousSibling.type === "chapter" &&
+                !error.children.some(ch => ch.type === "paragraph")) {
+                // console.log("Match 7.1");
+                const start = error.previousSibling.startIndex;
+                const end = error.startIndex;
+                const toReplace = modifiedUSFM.slice(start, end);
+                modifiedUSFM = modifiedUSFM.replace(toReplace, `${toReplace}\\p\n`);
+                changed = true;
+            }
             // Stray slash not with a valid marker
             else if (errorText.startsWith("\\") && !validMarkersPattern.test(errorText)) {
                 // console.log("Match 8");