DOC-5093 added TextPosInfo to hold shortcode position

Author: andy-stark-redis

build/image_report.py

@@ -4,6 +4,39 @@
 from pylibs.hugotools import ShortcodeIterator
 
 import argparse
+import os
+
+
+def scan_file(path: str) -> int:
+    """Scans a file for all `image` shortcodes.
+
+    Args:
+        path (str): Path to file.
+
+    Returns:
+        (int) Number of shortcodes found.
+    """
+
+    img_list = []
+
+    with open(path, encoding="utf_8") as md_file:
+        text = md_file.read()
+
+        for img, pos_info in ShortcodeIterator(
+            text, lambda t: t.tag == "image"
+        ):
+            img_list.append((img, pos_info))
+
+    if len(img_list) > 0:
+        print(f"File '{path}':")
+
+        for img in img_list:
+            print(
+                f"    Line {img[1].line}: '{img[0].named_params['filename']}'"
+            )
+
+    return len(img_list)
+
 
 parser = argparse.ArgumentParser(
     "Image report",
@@ -16,8 +49,15 @@
 
 print(f"Scanning '{args.pathname}'")
 
-with open(args.pathname, encoding="utf_8") as md_file:
-    filetext = md_file.read()
+num_found = 0
+
+for root, dirs, files in os.walk(args.pathname):
+    for file in files:
+        if file.endswith(".md"):
+            fullpath = os.path.join(root, file)
+            num_found += scan_file(fullpath)
 
-    for shortcode, line in ShortcodeIterator(filetext):
-        print(f"Line: {line}: {shortcode}")
+if num_found == 0:
+    print(f"No image shortcodes found in '{args.pathname}'")
+else:
+    print(f"Found {num_found} image shortcodes.")

build/pylibs/hugotools.py

@@ -3,6 +3,18 @@
 
 import re
 
+
+class TextPosInfo:
+    line: int
+    start: int
+    end: int
+
+    def __init__(self, line, start, end):
+        self.line = line
+        self.start = start
+        self.end = end
+
+
 shortcode_re_pattern_start = r"(\n)|\{\{[<%]\s*"
 shortcode_re_body = r"(/)?([\w\-]+)\s*(.+?)?"
 shortcode_re_pattern_end = r"\s*[>%]\}\}"
@@ -103,7 +115,7 @@ def __init__(
     def __iter__(self):
         return self
 
-    def __next__(self) -> tuple[ShortcodeInfo, int]:
+    def __next__(self) -> tuple[ShortcodeInfo, TextPosInfo]:
         next_match = self.re_iterator.__next__()
 
         while True:
@@ -115,7 +127,14 @@ def __next__(self) -> tuple[ShortcodeInfo, int]:
                 )
 
                 if self.sc_filter(self, result):
-                    return (result, self.linenum)
+                    return (
+                        result,
+                        TextPosInfo(
+                            self.linenum,
+                            next_match.start(),
+                            next_match.end()
+                        )
+                    )
             else:
                 result = ShortcodeInfo(
                     next_match[3],
@@ -124,6 +143,13 @@ def __next__(self) -> tuple[ShortcodeInfo, int]:
                 )
 
                 if self.sc_filter(self, result):
-                    return (result, self.linenum)
+                    return (
+                        result,
+                        TextPosInfo(
+                            self.linenum,
+                            next_match.start(),
+                            next_match.end()
+                        )
+                    )
 
             next_match = self.re_iterator.__next__()