From c87a50f36573d01e1eb1137ebdd4419d9f2930b3 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Sergio=20Mart=C3=ADnez=20Portela?= Date: Mon, 16 May 2022 23:16:20 +0200 Subject: [PATCH 1/2] Fix handling of Headline hierarchy. --- org_rw/org_rw.py | 17 ++++++++++++++--- 1 file changed, 14 insertions(+), 3 deletions(-) diff --git a/org_rw/org_rw.py b/org_rw/org_rw.py index c006d1d..cc4638f 100644 --- a/org_rw/org_rw.py +++ b/org_rw/org_rw.py @@ -1476,6 +1476,9 @@ def parse_headline(hl, doc, parent) -> Headline: contents = parse_contents(hl["contents"]) + if not (isinstance(parent, OrgDoc) or depth > parent.depth): + raise AssertionError("Incorrectly parsed parent on `{}' > `{}'".format(parent.title, title)) + headline = Headline( start_line=hl["linenum"], depth=depth, @@ -1781,18 +1784,26 @@ class OrgDocReader: "list_items": [], } - while (depth - 2) > len(self.headline_hierarchy): + while (depth - 1) > len(self.headline_hierarchy): # Introduce structural headlines self.headline_hierarchy.append(None) - while depth < len(self.headline_hierarchy): + while depth <= len(self.headline_hierarchy): self.headline_hierarchy.pop() if depth == 1: self.headlines.append(headline) else: - self.headline_hierarchy[-1]["children"].append(headline) + parent_idx = len(self.headline_hierarchy) - 1 + while self.headline_hierarchy[parent_idx] is None: + parent_idx -= 1 + self.headline_hierarchy[parent_idx]["children"].append(headline) self.headline_hierarchy.append(headline) + if all([hl is not None for hl in self.headline_hierarchy]): + if not ([ len(hl['orig'].group('stars')) for hl in self.headline_hierarchy ] + == list(range(1, len(self.headline_hierarchy) + 1))): + raise AssertionError('Error on Headline Hierarchy') + def add_list_item_line(self, linenum: int, match: re.Match) -> int: li = ListItem( linenum, From 63bb1e67e0e312c35f0d0c85269330da9f3e8d28 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Sergio=20Mart=C3=ADnez=20Portela?= Date: Mon, 16 May 2022 23:17:28 +0200 Subject: [PATCH 2/2] Handle :RESULTS: drawer, and data inside :LOGBOOK: --- org_rw/dom.py | 13 ++++++++++++- org_rw/org_rw.py | 27 ++++++++++++++++++++------- 2 files changed, 32 insertions(+), 8 deletions(-) diff --git a/org_rw/dom.py b/org_rw/dom.py index cda195c..6f0bce8 100644 --- a/org_rw/dom.py +++ b/org_rw/dom.py @@ -17,7 +17,18 @@ class LogbookDrawerNode: self.children.append(child) def __repr__(self): - return "".format(len(self.children)) + return "".format(len(self.children)) + + +class ResultsDrawerNode: + def __init__(self): + self.children = [] + + def append(self, child): + self.children.append(child) + + def __repr__(self): + return "".format(len(self.children)) class PropertyNode: diff --git a/org_rw/org_rw.py b/org_rw/org_rw.py index cc4638f..b4526be 100644 --- a/org_rw/org_rw.py +++ b/org_rw/org_rw.py @@ -98,7 +98,7 @@ RESULTS_DRAWER_RE = re.compile(r"^\s*:results:\s*$", re.I) CodeSnippet = collections.namedtuple("CodeSnippet", ("name", "content", "result")) # Groupings -NON_FINISHED_GROUPS = (type(None), dom.ListGroupNode) +NON_FINISHED_GROUPS = (type(None), dom.ListGroupNode, dom.ResultsDrawerNode, dom.PropertyDrawerNode) FREE_GROUPS = (dom.CodeBlock,) @@ -181,9 +181,9 @@ def text_to_dom(tokens, item): in_description = False link_value = [] link_description = [] - + contents = [] - + for tok in tokens: if isinstance(tok, LinkToken): if tok.tok_type == LinkTokenType.OPEN_LINK: @@ -362,7 +362,7 @@ class Headline: current_node.append(dom.Text(line)) else: if type(current_node) not in NON_FINISHED_GROUPS: - assert type(current_node) in NON_FINISHED_GROUPS + raise NotImplementedError('Not implemented node type: {}'.format(current_node)) current_node = None contents = [] tree.append(dom.Text(text_to_dom(line.contents, line))) @@ -373,7 +373,9 @@ class Headline: tree.append(current_node) indentation_tree = [current_node] if not isinstance(current_node, dom.ListGroupNode): - assert isinstance(current_node, dom.ListGroupNode) + if not isinstance(current_node, dom.ListGroupNode): + logging.warning("Expected a {}, found: {} on line {}".format(dom.ListGroupNode, current_node, line.linenum)) + # This can happen. Frequently inside a LogDrawer if len(indentation_tree) > 0 and ( (len(indentation_tree[-1].children) > 0) @@ -447,10 +449,21 @@ class Headline: current_node = dom.LogbookDrawerNode() tree.append(current_node) elif content.strip().upper() == ":END:": - assert isinstance( + if current_node is None: + logging.warning('Finished node (:END:) with no known starter') + elif not (isinstance( current_node, dom.PropertyDrawerNode - ) or isinstance(current_node, dom.LogbookDrawerNode) + ) or isinstance( + current_node, dom.LogbookDrawerNode + ) or isinstance( + current_node, dom.ResultsDrawerNode + )): + raise Exception('Unexpected node: {}'.format(current_node)) current_node = None + elif content.strip().upper() == ":RESULTS:": + assert current_node is None + current_node = dom.ResultsDrawerNode() + tree.append(current_node) else: raise Exception("Unknown structural line: {}".format(line)) else: