implements filenames and line numbers for syntax errors

Author: drhodes

src/auxml/macro.py

@@ -2,6 +2,8 @@
 from copy import deepcopy
 from auxml.util import *
 from auxml.err import SyntaxErrorAuXML
+import auxml.parser as parser
+
 '''
 <define-macro name="blue"><span style="color: #00f"><b><contents/></b></span></define-macro>
 '''
@@ -27,7 +29,22 @@ def append_tail(el, s):
 class MacroDef():
     def __init__(self, el):        
         self.el = el
+        self.check_for_name()
         self.name = el.get("name")
+
+    def check_for_name(self):
+        if self.el.get("name") is None:
+            fileinfo = parser.el_location_info(self.el)
+            msg = f'''
+            Encountered macro definition with missing `name` attribute
+            {fileinfo}
+            
+            Suggestion: change the macro definition include a name attribute:
+            
+            <define-macro name="..."> ... </define-macro>
+                          ^^^^
+            '''
+            raise SyntaxErrorAuXML(msg)
 
     def replace_one_var(self, el, varname, valuem):
         pass
@@ -39,12 +56,15 @@ def get_body(self):
         # the following line assumes the macro body only has one element.
         # TODO let macros definition have text and elements.
         cs = self.el.getchildren()
+        fileinfo = parser.el_location_info(self.el)
 
         if len(cs) == 0:
-            raise SyntaxErrorAuXML(f"Encountered empty macro body in macro definition: `{self.name}`")
+            # maybe this should be a warning?
+            msg = f"Encountered empty macro body in macro definition: `{self.name}`, {fileinfo}"
+            raise SyntaxErrorAuXML(msg)
 
         if len(cs) > 1:
-            msg = "macro definitions may not yet have more than one element"
+            msg = f"Macro definitions may not yet have more than one element, {fileinfo}"
             raise SyntaxErrorAuXML(msg)
 
         return deepcopy(self.el.getchildren()[0])
@@ -83,8 +103,9 @@ def attr_vars(self):
 
     def ensure_attrs_match(self, mcall):
         for var in self.attr_vars():
-            if not mcall.contains_attr(var):                
-                raise Exception(f"Macro call: {mcall.name()} on line ... must have attribute: {var}")
+            if not mcall.contains_attr(var):
+                info = mcall.fileinfo()
+                raise SyntaxErrorAuXML(f"Macro call: {mcall.name()} on line ... must have attribute: {var} {info}")
 
 
     def replace_one_content(self, mcall, con):
@@ -196,6 +217,9 @@ def __init__(self, el):
         self.el = el
         self.counter = MacroCall.counter
         MacroCall.counter += 1
+
+    def fileinfo(self):
+        return parser.el_location_info(self.el)        
 
     def unique_id(self):
         return f"{self.name()}-{self.counter}"

src/auxml/parser.py

@@ -1,15 +1,39 @@
 from lxml import etree
 from bs4 import BeautifulSoup
+import pudb
+
 _html_parser = etree.HTMLParser(remove_blank_text=True, remove_comments=True)
 _xml_parser = etree.XMLParser(remove_blank_text=True, remove_comments=True)
 
+ATTR_FILENAME = 'auxml-data-filename'
+ATTR_LINENUM = 'auxml-data-line'
+
+def el_location_info(el):
+    filename = el.get(ATTR_FILENAME)
+    linenum = el.get(ATTR_LINENUM)
+    return f'''
+    in file   ...... : {filename}
+    on line number.. : {linenum}
+    '''
+
+def tag_els_with_info(el, fname):
+    if hasattr(el, 'sourceline'): 
+        line_number = el.sourceline
+        el[ATTR_FILENAME] = fname
+        el[ATTR_LINENUM] = str(line_number)
+    
+    for child in el.children:
+        if isinstance(child, str):
+            continue        
+        tag_els_with_info(child, fname)
+
 def parse_html_file(fname):
     text = open(fname).read()
     soup = BeautifulSoup(text, "html.parser")
+    
+    # mutate soup in place
+    tag_els_with_info(soup, fname)
+    
     tree = etree.fromstring(str(soup), _xml_parser)
     return tree
 
-# def parse_xml_file(fname):
-#     tree = etree.parse(fname, _xml_parser)
-#     return tree.getroot()
-

tests/fileinfo_test.py

@@ -0,0 +1,22 @@
+import pytest
+import sys
+import pudb
+from lxml import etree
+from xmldiff.diff import Differ
+from auxml.macro import MacroDef, MacroCall
+from auxml.macro_manager import MacroManager
+from auxml.util import *
+from auxml.err import SyntaxErrorAuXML
+
+
+def with_mm_file(filename):
+    mm = MacroManager()
+    mm.load_macro_file(filename)
+    
+def causes_syntax_err(filename):
+    with pytest.raises(SyntaxErrorAuXML):
+        with_mm_file(filename)
+        
+def test_fileinfo():
+    causes_syntax_err("tests/test-html/macros-fileinfo-1.html")
+

tests/test-html/macros-fileinfo-1.html

@@ -0,0 +1,5 @@
+<macros>
+	<define-macro namename="bad">
+	    <div>this macro needs an attribute called "name"</div>   
+	</define-macro>
+</macros>