Use pathlib.

Read files as cp437 so all bytes are valid and windows-1252 invalid characters
do not fail.
When new headers found, save reasonable order of headers to "NewOrder.txt"
making it easier to update HeaderOrder.txt.
When a file is out-of-order, save list of headers used in the same order as
HeaderOrder.txt to "<path>.ordered".

1 files changed, 64 insertions, 28 deletions

diff --git a/scripts/HeaderCheck.py b/scripts/HeaderCheck.py
index 058832e31..29dc4fba3 100644
--- a/scripts/HeaderCheck.py
+++ b/scripts/HeaderCheck.py
@@ -3,7 +3,7 @@
 # Canonical header order is defined in scripts/HeaderOrder.txt
 # Requires Python 3.6 or later
 
-import codecs, glob, os
+import pathlib
 
 patterns = [
     "include/*.h",
@@ -20,65 +20,101 @@ patterns = [
 ]
 
 def IsHeader(x):
-    return x.strip().startswith("#") and ("include" in x or "import" in x)
+    return x.strip().startswith("#") and \
+        ("include" in x or "import" in x) and \
+        "dllimport" not in x
 
 def HeaderFromIncludeLine(s):
     #\s*#\s*(include|import)\s+\S+\s*
     return s.strip()[1:].strip()[7:].strip()
 
 def ExtractHeaders(filename):
-    with codecs.open(filename, "r", "windows-1252") as infile:
+    with filename.open(encoding="cp437") as infile:
         return [HeaderFromIncludeLine(l) for l in infile if IsHeader(l)]
 
+def ExcludeName(name):
+    # LexCaml adds system headers in #if to be an external lexer
+    # moc_ files are generated by Qt and follow its rules
+    return "LexCaml" in name or "moc_" in name
+
+def SortLike(incs, order):
+    return sorted(incs, key = lambda i: order.index(i))
+
 def CheckFiles(root):
     # Find all the lexer source code files
     filePaths = []
     for p in patterns:
-        filePaths += glob.glob(os.path.join(root, p))
+        filePaths += root.glob(p)
     # The Qt platform code interleaves system and Scintilla headers
-    #~ filePaths += glob.glob(root + "/qt/ScintillaEditBase/*.cpp")
-    #~ filePaths += glob.glob(root + "/qt/ScintillaEdit/*.cpp")
+    #~ filePaths += root.glob("qt/ScintillaEditBase/*.cpp")
+    #~ filePaths += root.glob("qt/ScintillaEdit/*.cpp")
     #~ print(filePaths)
-    masterHeaderList = ExtractHeaders(os.path.join(root, "scripts/HeaderOrder.txt"))
-    orderedPaths = sorted(filePaths, key=str.casefold)
+    scriptDirectory = root / "scripts"
+    headerOrderFile = scriptDirectory / "HeaderOrder.txt"
+    headerOrder = ExtractHeaders(headerOrderFile)
+    originalOrder = headerOrder[:]
+    orderedPaths = [p for p in sorted(filePaths) if not ExcludeName(str(p))]
     allIncs = set()
     for f in orderedPaths:
-        if "LexCaml" in f: # LexCaml adds system headers in #if to be an external lexer
-            continue
-        print("   File ", f)
+        print("   File ", f.relative_to(root))
         incs = ExtractHeaders(f)
         #~ print("\n".join(incs))
-        news = set(incs) - set(masterHeaderList)
+        news = set(incs) - set(headerOrder)
         allIncs = allIncs.union(set(incs))
+
         m = 0
         i = 0
+        # Detect headers not in header order list and insert at OK position
+        needs = []
         while i < len(incs):
-            if m == len(masterHeaderList):
-                print("**** extend", incs[i:])
-                masterHeaderList.extend(incs[i:])
+            if m == len(headerOrder):
+                #~ print("**** extend", incs[i:])
+                headerOrder.extend(incs[i:])
+                needs.extend(incs[i:])
                 break
-            if masterHeaderList[m] == incs[i]:
-                #~ print("equal", masterHeaderList[m])
+            if headerOrder[m] == incs[i]:
+                #~ print("equal", headerOrder[m])
                 i += 1
                 m += 1
             else:
-                if masterHeaderList[m] not in incs:
-                    #~ print("skip", masterHeaderList[m])
+                if headerOrder[m] not in incs:
+                    #~ print("skip", headerOrder[m])
                     m += 1
-                elif incs[i] not in masterHeaderList:
-                    print(f + ":1: Add master", incs[i])
-                    masterHeaderList.insert(m, incs[i])
+                elif incs[i] not in headerOrder:
+                    #~ print(str(f) + ":1: Add master", incs[i])
+                    headerOrder.insert(m, incs[i])
+                    needs.append(incs[i])
                     i += 1
                     m += 1
                 else:
-                    print(f + ":1: Header out of order", incs[i], masterHeaderList[m])
-                    print("incs", " ".join(incs))
                     i += 1
-                    #~ return
-        #print("Master header list", " ".join(masterHeaderList))
-    unused = sorted(set(masterHeaderList) - allIncs)
+        if needs:
+            print(f"{f}:1: needs these headers:")
+            for header in needs:
+                print("#include " + header)
+
+        # Detect out of order
+        ordered = SortLike(incs, headerOrder)
+        if incs != ordered:
+            print(f"{f}:1: is out of order")
+            fOrdered = pathlib.Path(str(f) + ".ordered")
+            with fOrdered.open("w") as headerOut:
+                for header in ordered:
+                    headerOut.write("#include " + header + "\n")
+            print(f"{fOrdered}:1: is ordered")
+
+    if headerOrder != originalOrder:
+        newIncludes = set(headerOrder) - set(originalOrder)
+        headerOrderNew = scriptDirectory / "NewOrder.txt"
+        print(f"{headerOrderFile}:1: changed to {headerOrderNew}")
+        print(f"   Added {', '.join(newIncludes)}.")
+        with headerOrderNew.open("w") as headerOut:
+            for header in headerOrder:
+                headerOut.write("#include " + header + "\n")
+
+    unused = sorted(set(headerOrder) - allIncs)
     if unused:
         print("In HeaderOrder.txt but not used")
         print("\n".join(unused))
 
-CheckFiles("..")
+CheckFiles(pathlib.Path(__file__).resolve().parent.parent)