Less back and forths by returning semi count in page

aothms · aothms · commit e82ca0e93211 · 2025-10-25T15:08:21.000+02:00
diff --git a/src/ifcopenshell-python/ifcopenshell/geom/stats.py b/src/ifcopenshell-python/ifcopenshell/geom/stats.py
@@ -138,6 +138,8 @@ class StatsCollector:
 
     counters: list
 
+    num_semis: int = 0
+
     def __init__(self):
         self.streamer = ifcopenshell_wrapper.InstanceStreamer()
         self.needs_data = True
@@ -151,17 +153,19 @@ def feedFromFile(self, f: Optional[IO[str]] = None):
     def feed(self, data: str):
         self.streamer.pushPage(data)
         self.needs_data = False
+        self.num_semis = self.streamer.semicolonCount()
 
     @staticmethod
-    def fromFilePath(fn, page_size: int = 4096):
+    def fromFilePath(fn, page_size: int = 102400):
         collector = StatsCollector()
         collector.page_size = page_size
         collector.feedFromFile(open(str(fn), encoding="ascii"))
         return collector
 
     def next(self):
-        if self.streamer.hasSemicolon():
+        if self.num_semis > 0:
             if inst := self.streamer.readInstancePy(True):
+                self.num_semis -= 1
                 return inst["type"], dict(list(inst.items())[2:])
             else:
                 self.finalized = True
diff --git a/src/ifcparse/IfcFile.h b/src/ifcparse/IfcFile.h
@@ -139,7 +139,9 @@ class IFC_PARSE_API InstanceStreamer {
 
     bool hasSemicolon() const;
 
-	void pushPage(const std::string& page);
+    size_t semicolonCount() const;
+
+    void pushPage(const std::string& page);
 
     InstanceStreamer();
 
diff --git a/src/ifcparse/IfcParse.cpp b/src/ifcparse/IfcParse.cpp
@@ -1371,6 +1371,30 @@ bool IfcParse::InstanceStreamer::hasSemicolon() const {
 	return false;
 }
 
+size_t IfcParse::InstanceStreamer::semicolonCount() const {
+    auto local_stream = stream_->clone();
+    auto local_lexer = IfcSpfLexer(&local_stream);
+    Token t;
+    size_t count = 0;
+    try {
+        t = local_lexer.Next();
+    } catch (const std::out_of_range&) {
+        return false;
+    }
+    while (t.type != Token_NONE) {
+        if (TokenFunc::isOperator(t, ';')) {
+            count++;
+        }
+        try {
+            t = local_lexer.Next();
+        } catch (const std::out_of_range&) {
+            // This most likely happens when a page boundary is contained within a string
+            break;
+        }
+    }
+    return count;
+}
+
 void IfcParse::InstanceStreamer::pushPage(const std::string& page)
 {
     stream_->pushNextPage(page);
