tools: rework generating documentation from libnm meta data

With the given input, this produces *exactly* the same XML as before.

- the parsing is now stricter (and thus the code more verbose). No
  funny stuff, get the annotations correct.
- on parsing errors, we log now the affecting lines
- "nm-setting-ip-config.c" is a base class. Previously it was ignored
  and for the moment we still do that. Next, we will allow to also
  describe properties there.
- prepare the code to better preserve whitespace, indentation and line
  wrappings. In particular, to honor a blank line to indicate a line break
  and support paragraphs. This is not yet done to compare the output
  to before, but will be turned on with a small patch next.
- the code will make it simple to promote the XML attributes to
  nodes. Attributes aren't great, let's write XML nodes later.
  We will only need to adjust the "keywords" dictionary for that,
  but this change will require changes to the entire chain of tools.

1 files changed, 382 insertions, 113 deletions

diff --git a/tools/generate-docs-nm-property-infos.py b/tools/generate-docs-nm-property-infos.py
index 25aa272a69..568bab17bc 100755
--- a/tools/generate-docs-nm-property-infos.py
+++ b/tools/generate-docs-nm-property-infos.py
@@ -1,143 +1,412 @@
 #!/usr/bin/env python
 # SPDX-License-Identifier: LGPL-2.1-or-later
 
+import os
 import re
 import sys
+import collections
 import xml.etree.ElementTree as ET
 
 
-def get_setting_name(one_file):
-    setting_name = ""
-    assert re.match(r".*/libnm-core-impl/nm-setting-.*\.c$", one_file)
-    header_path = one_file.replace("libnm-core-impl", "libnm-core-public")
-    header_path = header_path.replace(".c", ".h")
-    try:
-        header_reader = open(header_path, "r")
-    except OSError:
-        print("Can not open header file: %s" % (header_path))
-        exit(1)
+class LineError(Exception):
+    def __init__(self, line_no, msg):
+        Exception.__init__(self, msg)
+        self.line_no = line_no
+
+
+_dbg_level = 0
+try:
+    _dbg_level = int(os.getenv("NM_DEBUG_GENERATE_DOCS", 0))
+except Exception:
+    pass
+
+
+def dbg(msg, level=1):
+    if level <= _dbg_level:
+        print(msg)
+
+
+def iter_unique(iterable, default=None):
+    found = False
+    for i in iterable:
+        assert not found
+        found = True
+        i0 = i
+    if found:
+        return i0
+    return default
+
 
-    line = header_reader.readline()
-    while line != "":
-        setting_name_found = re.search(r"NM_SETTING_.+SETTING_NAME\s+\"(\S+)\"", line)
-        if setting_name_found:
-            setting_name = setting_name_found.group(1)
-            break
-        line = header_reader.readline()
-    header_reader.close()
-    return setting_name
+def xnode_get_or_create(root_node, node_name, name):
+    # From root_node, get the node "<{node_name} name={name} .../>"
+    # or create one, if it doesn't exist.
+    node = iter_unique(
+        (node for node in root_node.findall(node_name) if node.attrib["name"] == name)
+    )
+    if node is None:
+        created = True
+        node = ET.SubElement(root_node, node_name, name=name)
+    else:
+        created = False
 
+    return node, created
+
+
+def get_setting_names(source_file):
+    m = re.match(r"^(.*)/libnm-core-impl/(nm-setting-[^/]*)\.c$", source_file)
+    assert m
+
+    path_prefix, file_base = (m.group(1), m.group(2))
+
+    if file_base == "nm-setting-ip-config":
+        # Special case ip-config, which is a base class.
+        return None
+
+    header_file = "%s/libnm-core-public/%s.h" % (path_prefix, file_base)
 
-def scan_doc_comments(plugin, setting_node, file, start_tag, end_tag):
-    data = []
-    push_flag = 0
     try:
-        file_reader = open(file, "r")
+        f = open(header_file, "r")
     except OSError:
-        print("Can not open file: %s" % (file))
-        exit(1)
+        raise Exception(
+            'Can not open header file "%s" for "%s"' % (header_file, source_file)
+        )
+
+    with f:
+        for line in f:
+            m = re.search(r"^#define +NM_SETTING_.+SETTING_NAME\s+\"(\S+)\"$", line)
+            if m:
+                return m.group(1)
+
+    raise Exception(
+        'Can\'t find setting name in header file "%s" for "%s"'
+        % (header_file, source_file)
+    )
 
-    line = file_reader.readline()
-    while line != "":
-        if start_tag in line:
-            push_flag = 1
-        elif end_tag in line and push_flag == 1:
-            push_flag = 0
-            parsed_data = process_data(data)
-            if parsed_data:
-                write_data(setting_node, parsed_data)
-            data = []
-        elif push_flag == 1:
-            data.append(line)
-        line = file_reader.readline()
-    file_reader.close()
-    return
-
-
-keywords = [
-    "property",
-    "variable",
-    "format",
-    "values",
-    "default",
-    "example",
-    "description",
-    "description-docbook",
-]
-kwd_first_line_re = re.compile(
-    r"^\s*\**\s+({}):\s+(.*?)\s*$".format("|".join(keywords))
+
+def get_file_infos(source_files):
+    for source_file in source_files:
+        setting_name = get_setting_names(source_file)
+        if setting_name:
+            yield setting_name, source_file
+
+
+KEYWORD_XML_TYPE_NESTED = "nested"
+KEYWORD_XML_TYPE_NODE = "node"
+KEYWORD_XML_TYPE_ATTR = "attr"
+
+keywords = collections.OrderedDict(
+    [
+        ("property", KEYWORD_XML_TYPE_ATTR),
+        ("variable", KEYWORD_XML_TYPE_ATTR),
+        ("format", KEYWORD_XML_TYPE_ATTR),
+        ("values", KEYWORD_XML_TYPE_ATTR),
+        ("default", KEYWORD_XML_TYPE_ATTR),
+        ("example", KEYWORD_XML_TYPE_ATTR),
+        ("description", KEYWORD_XML_TYPE_ATTR),
+        ("description-docbook", KEYWORD_XML_TYPE_NESTED),
+    ]
 )
-kwd_more_line_re = re.compile(r"^\s*\**\s+(.*?)\s*$")
 
 
-def process_data(data):
+def keywords_allowed(tag, keyword):
+    # certain keywords might not be valid for some tags.
+    # Currently, all of them are always valid.
+    assert keyword in keywords
+    return True
+
+
+def write_data(tag, setting_node, line_no, parsed_data):
+
+    for k in parsed_data.keys():
+        assert keywords_allowed(tag, k)
+        assert k in keywords
+
+    name = parsed_data["property"]
+    property_node, created = xnode_get_or_create(setting_node, "property", name)
+    if not created:
+        raise LineError(line_no, 'Duplicate property <property name="%s"...' % (name,))
+
+    for k, xmltype in keywords.items():
+        if k == "property":
+            continue
+
+        v = parsed_data.get(k, None)
+        if v is None:
+            if k == "variable":
+                v = name
+            elif k == 'description-docbook':
+                continue
+            else:
+                v = ""
+
+        if xmltype == KEYWORD_XML_TYPE_NESTED:
+            # Set as XML nodes. The input data is XML itself.
+            des = ET.fromstring("<%s>%s</%s>" % (k, v, k))
+            property_node.append(des)
+        elif xmltype == KEYWORD_XML_TYPE_NODE:
+            node = ET.SubElement(property_node, k)
+            node.text = v
+        elif xmltype == KEYWORD_XML_TYPE_ATTR:
+            property_node.set(k, v)
+        else:
+            assert False
+
+
+kwd_first_line_re = re.compile(r"^ *\* ([-a-z0-9]+): (.*)$")
+kwd_more_line_re = re.compile(r"^ *\*( *)(.*?)\s*$")
+
+
+def parse_data(tag, line_no, lines):
+    assert lines
     parsed_data = {}
-    if not data:
-        return parsed_data
     keyword = ""
-    for line in data:
-        kwd_first_line_found = kwd_first_line_re.search(line)
-        if kwd_first_line_found:
-            keyword = kwd_first_line_found.group(1)
-            if keyword == "description-docbook":
-                value = kwd_first_line_found.group(2) + "\n"
+    first_line = True
+    indent = None
+    for line in lines:
+        assert "\n" not in line
+        line_no += 1
+        m = re.search(r"^     \*(| .*)$", line)
+        if not m:
+            raise LineError(line_no, 'Invalid formatted line "%s"' % (line,))
+        content = m.group(1)
+
+        m = re.search("^ ([-a-z0-9]+):(.*)$", content)
+        text_keyword_started = None
+        if m:
+            keyword = m.group(1)
+            if keyword in parsed_data:
+                raise LineError(line_no, 'Duplicated keyword "%s"' % (keyword,))
+            text = m.group(2)
+            text_keyword_started = text
+            if text:
+                if text[0] != " " or len(text) == 1:
+                    raise LineError(line_no, 'Invalid formatted line "%s"' % (line,))
+                text = text[1:]
+            if not keywords_allowed(tag, keyword):
+                raise LineError(line_no, 'Invalid key "%s" for %s' % (keyword, tag))
+            if parsed_data and keyword == "property":
+                raise LineError(line_no, 'The "property:" keywork must be first')
+            parsed_data[keyword] = text
+            new_keyword_stated = True
+            indent = None
+        else:
+            if content == "":
+                text = ""
+            elif content[0] == " " and len(content) > 1:
+                text = content[1:]
+                assert text
+                if indent is None:
+                    indent = re.search("^( *)", text).group(1)
+                if not text.startswith(indent):
+                    raise LineError(line_no, 'Unexpected indention in "%s"' % (line,))
+                text = text[len(indent) :]
             else:
-                value = kwd_first_line_found.group(2) + " "
-            parsed_data[keyword] = value
-            continue
-        kwd_more_line_found = kwd_more_line_re.search(line)
-        if kwd_more_line_found:
+                raise LineError(line_no, 'Unexpected line "%s"' % (line,))
             if not keyword:
-                print("Extra mess in a comment: %s" % (line))
-                exit(1)
-            if keyword == "description-docbook":
-                value = kwd_more_line_found.group(1) + "\n"
+                raise LineError(line_no, "Expected data in comment: %s" % (line))
+            if text and text[0] == "\\":
+                assert False
+                text = text[1:]
+            if separator == " " and text == "":
+                # No separator to add. This is a blank line
+                pass
             else:
-                value = kwd_more_line_found.group(1) + " "
-            parsed_data[keyword] += value
-    for keyword in keywords:
-        if keyword == "variable" and keyword not in parsed_data:
-            parsed_data[keyword] = parsed_data["property"]
-        elif keyword not in parsed_data:
-            parsed_data[keyword] = ""
-    for key in parsed_data.keys():
-        parsed_data[key] = parsed_data[key].rstrip()
+                parsed_data[keyword] = parsed_data[keyword] + separator + text.strip()
+
+        if keywords[keyword] == KEYWORD_XML_TYPE_NESTED:
+            # This is plain XML. They lines are joined by newlines.
+            separator = "\n"
+        elif text_keyword_started == "":
+            # If the previous line was just "tag:$", we don't need a separator
+            # the next time.
+            separator = ""
+        elif not text:
+            # A blank line is used to mark a line break, while otherwise
+            # lines are joined by space.
+            separator = "  "
+        else:
+            separator = " "
+    if "property" not in parsed_data:
+        raise LineError(line_no, 'Missing "property:" tag')
+    for keyword in keywords.keys():
+        if not keywords_allowed(tag, keyword):
+            continue
+        if keyword not in parsed_data:
+            parsed_data[keyword] = None
     return parsed_data
 
 
-def write_data(setting_node, parsed_data):
-    property_node = ET.SubElement(setting_node, "property")
-    property_node.set("name", parsed_data["property"])
-    property_node.set("variable", parsed_data["variable"])
-    property_node.set("format", parsed_data["format"])
-    property_node.set("values", parsed_data["values"])
-    property_node.set("default", parsed_data["default"])
-    property_node.set("example", parsed_data["example"])
-    property_node.set("description", parsed_data["description"])
-    if parsed_data["description-docbook"]:
-        des = ET.fromstring(
-            "<description-docbook>"
-            + parsed_data["description-docbook"]
-            + "</description-docbook>"
-        )
-        property_node.append(des)
+def process_setting(tag, root_node, source_file, setting_name):
 
+    dbg(
+        "> > tag:%s, source_file:%s, setting_name:%s" % (tag, source_file, setting_name)
+    )
 
-if len(sys.argv) < 4:
-    print("Usage: %s [plugin] [output-xml-file] [srcfiles]" % (sys.argv[0]))
-    exit(1)
+    start_tag = "---" + tag + "---"
+    end_tag = "---end---"
 
-argv = list(sys.argv[1:])
-plugin, output, source_files = argv[0], argv[1], argv[2:]
-start_tag = "---" + plugin + "---"
-end_tag = "---end---"
-root_node = ET.Element("nm-setting-docs")
-
-for one_file in source_files:
-    setting_name = get_setting_name(one_file)
-    if setting_name:
-        setting_node = ET.SubElement(root_node, "setting", name=setting_name)
+    setting_node, created = xnode_get_or_create(root_node, "setting", setting_name)
+    if created:
         setting_node.text = "\n"
-        scan_doc_comments(plugin, setting_node, one_file, start_tag, end_tag)
 
-ET.ElementTree(root_node).write(output)
+    try:
+        f = open(source_file, "r")
+    except OSError:
+        raise Exception("Can not open file: %s" % (source_file))
+
+    lines = None
+    with f:
+        line_no = 0
+        just_had_end_tag = False
+        line_no_start = None
+        for line in f:
+            line_no += 1
+            if line and line[-1] == "\n":
+                line = line[:-1]
+            if just_had_end_tag:
+                # After the end-tag, we still expect one particular line. Be strict about
+                # this.
+                just_had_end_tag = False
+                if line != "     */":
+                    raise LineError(
+                        line_no,
+                        'Invalid end tag "%s". Expects literally "     */" after end-tag'
+                        % (line,),
+                    )
+            elif start_tag in line:
+                if line != "    /* " + start_tag:
+                    raise LineError(
+                        line_no,
+                        'Invalid start tag "%s". Expects literally "    /* %s"'
+                        % (line, start_tag),
+                    )
+                if lines is not None:
+                    raise LineError(
+                        line_no, 'Invalid start tag "%s", missing end-tag' % (line,)
+                    )
+                lines = []
+                line_no_start = line_no
+            elif end_tag in line and lines is not None:
+                if line != "     * " + end_tag:
+                    raise LineError(line_no, 'Invalid end tag: "%s"' % (line,))
+                parsed_data = parse_data(tag, line_no_start, lines)
+                if not parsed_data:
+                    raise Exception('invalid data: line %s, "%s"' % (line_no, lines))
+                dbg("> > > property: %s" % (parsed_data["property"],))
+                if _dbg_level > 1:
+                    for keyword in sorted(parsed_data.keys()):
+                        v = parsed_data[keyword]
+                        if v is not None:
+                            v = '"%s"' % (v,)
+                        dbg(
+                            "> > > > [%s] (%s) = %s" % (keyword, keywords[keyword], v),
+                            level=2,
+                        )
+                write_data(tag, setting_node, line_no_start, parsed_data)
+                lines = None
+            elif lines is not None:
+                lines.append(line)
+        if lines is not None or just_had_end_tag:
+            raise LineError(line_no_start, "Unterminated start tag")
+
+
+def process_settings_docs(tag, output, source_files):
+
+    dbg("> tag:%s, output:%s" % (tag, output))
+
+    root_node = ET.Element("nm-setting-docs")
+
+    for setting_name, source_file in get_file_infos(source_files):
+        try:
+            process_setting(tag, root_node, source_file, setting_name)
+        except LineError as e:
+            raise Exception(
+                "Error parsing %s, line %s (tag:%s, setting_name:%s): %s"
+                % (source_file, e.line_no, tag, setting_name, str(e))
+            )
+        except Exception as e:
+            raise Exception(
+                "Error parsing %s (tag:%s, setting_name:%s): %s"
+                % (source_file, tag, setting_name, str(e))
+            )
+
+    ET.ElementTree(root_node).write(output)
+
+
+def main():
+    if len(sys.argv) < 4:
+        print("Usage: %s [tag] [output-xml-file] [srcfiles...]" % (sys.argv[0]))
+        exit(1)
+
+    process_settings_docs(
+        tag=sys.argv[1], output=sys.argv[2], source_files=sys.argv[3:]
+    )
+
+
+if __name__ == "__main__":
+    main()
+
+
+###############################################################################
+# Tests
+###############################################################################
+
+
+def setup_module():
+    global pytest
+    import pytest
+
+
+def t_srcdir():
+    return os.path.abspath(os.path.dirname(__file__) + "/..")
+
+
+def t_setting_c(name):
+    return t_srcdir() + f"/src/libnm-core-impl/nm-setting-{name}.c"
+
+
+def test_file_location():
+    assert t_srcdir() + "/tools/generate-docs-nm-property-infos.py" == os.path.abspath(
+        __file__
+    )
+    assert os.path.isfile(t_srcdir() + "/src/libnm-core-impl/nm-setting-connection.c")
+
+    assert os.path.isfile(t_setting_c("ip-config"))
+
+
+def test_get_setting_names():
+    assert "connection" == get_setting_names(
+        t_srcdir() + "/src/libnm-core-impl/nm-setting-connection.c"
+    )
+    assert "ipv4" == get_setting_names(
+        t_srcdir() + "/src/libnm-core-impl/nm-setting-ip4-config.c"
+    )
+    assert None == get_setting_names(
+        t_srcdir() + "/src/libnm-core-impl/nm-setting-ip-config.c"
+    )
+
+
+def test_get_file_infos():
+
+    t = ["connection", "ip-config", "ip4-config", "proxy", "wired"]
+
+    assert [
+        (
+            "connection",
+            t_setting_c("connection"),
+        ),
+        (
+            "ipv4",
+            t_setting_c("ip4-config"),
+        ),
+        ("proxy", t_setting_c("proxy")),
+        (
+            "802-3-ethernet",
+            t_setting_c("wired"),
+        ),
+    ] == list(get_file_infos([t_setting_c(x) for x in t]))
+
+
+def test_process_setting():
+    root_node = ET.Element("nm-setting-docs")
+    process_setting("nmcli", root_node, t_setting_c("connection"), "connection")