klbr/khoj
1
0
Fork 0
mirror of https://github.com/khoaliber/khoj.git synced 2026-03-04 05:39:06 +00:00
Code Issues Packages Projects Releases Wiki Activity

Handle Empty Org Files or Org Files with No Headings

Browse Source 
### Main Changes
- bf01a4f Use filename or "#+TITLE" as heading for 0th level content in org files
- d6bd7bf Fix initializing `OrgNode` `level` to string to parse org files with no headings
- d835467 Throw exception if no valid entries found in specified content files

### Miscellaneous Improvements
- 7df39e5 Reuse search models across `pytest` sessions. Merge unused pytest fixtures
- 2dc0588 Do not normalize absolute filenames for entry links in `OrgNode`
- e00bb53 Init word filter dictionary with default value as set to simplify code

Resolves #83
This commit is contained in:
Debanjum
2022-09-10 12:42:07 +00:00
committed by GitHub
parent 4d776d9c7a 976397bd82
commit 372dcd2dbc
7 changed files with 190 additions and 50 deletions
Download Patch File Download Diff File Expand all files Collapse all files

32
src/processor/org_mode/orgnode.py
View File

@@ -41,8 +41,13 @@ from os.path import relpath
indent_regex = re.compile(r'^\s*')
def normalize_filename(filename):
file_relative_to_home = f'~/{relpath(filename, start=Path.home())}'
escaped_filename = f'{file_relative_to_home}'.replace("[","\[").replace("]","\]")
"Normalize and escape filename for rendering"
if not Path(filename).is_absolute():
# Normalize relative filename to be relative to current directory
normalized_filename = f'~/{relpath(filename, start=Path.home())}'
else:
normalized_filename = filename
escaped_filename = f'{normalized_filename}'.replace("[","\[").replace("]","\]")
return escaped_filename
def makelist(filename):
@@ -61,7 +66,7 @@ def makelist(filename):
todos = { "TODO": "", "WAITING": "", "ACTIVE": "",
"DONE": "", "CANCELLED": "", "FAILED": ""} # populated from #+SEQ_TODO line
level = 0
level = ""
heading = ""
bodytext = ""
tags = set() # set of all tags in headline
@@ -73,6 +78,7 @@ def makelist(filename):
propdict = dict()
in_properties_drawer = False
in_logbook_drawer = False
file_title = f'{filename}'
for line in f:
ctr += 1
@@ -111,6 +117,16 @@ def makelist(filename):
kwlist = re.findall(r'([A-Z]+)\(', line)
for kw in kwlist: todos[kw] = ""
# Set file title to TITLE property, if it exists
title_search = re.search(r'^#\+TITLE:\s*(.*)$', line)
if title_search and title_search.group(1).strip() != '':
title_text = title_search.group(1).strip()
if file_title == f'{filename}':
file_title = title_text
else:
file_title += f' {title_text}'
continue
# Ignore Properties Drawers Completely
if re.search(':PROPERTIES:', line):
in_properties_drawer=True
@@ -167,7 +183,7 @@ def makelist(filename):
bodytext = bodytext + line
# write out last node
thisNode = Orgnode(level, heading, bodytext, tags)
thisNode = Orgnode(level, heading or file_title, bodytext, tags)
thisNode.setProperties(propdict)
if sched_date:
thisNode.setScheduled(sched_date)
@@ -196,8 +212,12 @@ def makelist(filename):
n.setHeading(prtysrch.group(2))
# Set SOURCE property to a file+heading based org-mode link to the entry
escaped_heading = n.Heading().replace("[","\\[").replace("]","\\]")
n.properties['SOURCE'] = f'[[file:{normalize_filename(filename)}::*{escaped_heading}]]'
if n.Level() == 0:
n.properties['LINE'] = f'file:{normalize_filename(filename)}::0'
n.properties['SOURCE'] = f'[[file:{normalize_filename(filename)}]]'
else:
escaped_heading = n.Heading().replace("[","\\[").replace("]","\\]")
n.properties['SOURCE'] = f'[[file:{normalize_filename(filename)}::*{escaped_heading}]]'
return nodelist

6
src/search_filter/word_filter.py
View File

@@ -3,6 +3,7 @@ import re
import time
import pickle
import logging
from collections import defaultdict
# Internal Packages
from src.search_filter.base_filter import BaseFilter
@@ -37,19 +38,18 @@ class WordFilter(BaseFilter):
start = time.time()
self.cache = {} # Clear cache on (re-)generating entries_by_word_set
entry_splitter = r',|\.| |\]|\[\(|\)|\{|\}|\t|\n|\:'
self.word_to_entry_index = defaultdict(set)
# Create map of words to entries they exist in
for entry_index, entry in enumerate(entries):
for word in re.split(entry_splitter, entry[self.entry_key].lower()):
if word == '':
continue
if word not in self.word_to_entry_index:
self.word_to_entry_index[word] = set()
self.word_to_entry_index[word].add(entry_index)
with self.filter_file.open('wb') as f:
pickle.dump(self.word_to_entry_index, f)
end = time.time()
logger.debug(f"Index {self.search_type} for word filter to {self.filter_file}: {end - start} seconds")
logger.debug(f"Indexed {len(self.word_to_entry_index)} words of {self.search_type} type for word filter to {self.filter_file}: {end - start} seconds")
return self.word_to_entry_index

4
src/search_type/text_search.py
View File

@@ -11,7 +11,7 @@ from src.search_filter.base_filter import BaseFilter
# Internal Packages
from src.utils import state
from src.utils.helpers import get_absolute_path, resolve_absolute_path, load_model
from src.utils.helpers import get_absolute_path, is_none_or_empty, resolve_absolute_path, load_model
from src.utils.config import TextSearchModel
from src.utils.rawconfig import TextSearchConfig, TextContentConfig
from src.utils.jsonl import load_jsonl
@@ -174,6 +174,8 @@ def setup(text_to_jsonl, config: TextContentConfig, search_config: TextSearchCon
# Extract Entries
entries = extract_entries(config.compressed_jsonl)
if is_none_or_empty(entries):
raise ValueError(f"No valid entries found in specified files: {config.input_files} or {config.input_filter}")
top_k = min(len(entries), top_k) # top_k hits can't be more than the total entries in corpus
# Compute or Load Embeddings

45
tests/conftest.py
View File

@@ -4,6 +4,7 @@ import pytest
# Internal Packages
from src.search_type import image_search, text_search
from src.utils.config import SearchType
from src.utils.helpers import resolve_absolute_path
from src.utils.rawconfig import ContentConfig, TextContentConfig, ImageContentConfig, SearchConfig, TextSearchConfig, ImageSearchConfig
from src.processor.org_mode.org_to_jsonl import org_to_jsonl
from src.search_filter.date_filter import DateFilter
@@ -12,41 +13,41 @@ from src.search_filter.file_filter import FileFilter
@pytest.fixture(scope='session')
def search_config(tmp_path_factory) -> SearchConfig:
model_dir = tmp_path_factory.mktemp('data')
def search_config() -> SearchConfig:
model_dir = resolve_absolute_path('~/.khoj/search')
model_dir.mkdir(parents=True, exist_ok=True)
search_config = SearchConfig()
search_config.symmetric = TextSearchConfig(
encoder = "sentence-transformers/all-MiniLM-L6-v2",
cross_encoder = "cross-encoder/ms-marco-MiniLM-L-6-v2",
model_directory = model_dir
model_directory = model_dir / 'symmetric/'
)
search_config.asymmetric = TextSearchConfig(
encoder = "sentence-transformers/multi-qa-MiniLM-L6-cos-v1",
cross_encoder = "cross-encoder/ms-marco-MiniLM-L-6-v2",
model_directory = model_dir
model_directory = model_dir / 'asymmetric/'
)
search_config.image = ImageSearchConfig(
encoder = "sentence-transformers/clip-ViT-B-32",
model_directory = model_dir
model_directory = model_dir / 'image/'
)
return search_config
@pytest.fixture(scope='session')
def model_dir(search_config: SearchConfig):
model_dir = search_config.asymmetric.model_directory
def content_config(tmp_path_factory, search_config: SearchConfig):
content_dir = tmp_path_factory.mktemp('content')
# Generate Image Embeddings from Test Images
content_config = ContentConfig()
content_config.image = ImageContentConfig(
input_directories = ['tests/data/images'],
embeddings_file = model_dir.joinpath('image_embeddings.pt'),
batch_size = 10,
embeddings_file = content_dir.joinpath('image_embeddings.pt'),
batch_size = 1,
use_xmp_metadata = False)
image_search.setup(content_config.image, search_config.image, regenerate=False)
@@ -55,28 +56,10 @@ def model_dir(search_config: SearchConfig):
content_config.org = TextContentConfig(
input_files = None,
input_filter = 'tests/data/org/*.org',
compressed_jsonl = model_dir.joinpath('notes.jsonl.gz'),
embeddings_file = model_dir.joinpath('note_embeddings.pt'))
compressed_jsonl = content_dir.joinpath('notes.jsonl.gz'),
embeddings_file = content_dir.joinpath('note_embeddings.pt'))
filters = [DateFilter(), WordFilter(model_dir, search_type=SearchType.Org), FileFilter()]
filters = [DateFilter(), WordFilter(content_dir, search_type=SearchType.Org), FileFilter()]
text_search.setup(org_to_jsonl, content_config.org, search_config.asymmetric, regenerate=False, filters=filters)
return model_dir
@pytest.fixture(scope='session')
def content_config(model_dir) -> ContentConfig:
content_config = ContentConfig()
content_config.org = TextContentConfig(
input_files = None,
input_filter = 'tests/data/org/*.org',
compressed_jsonl = model_dir.joinpath('notes.jsonl.gz'),
embeddings_file = model_dir.joinpath('note_embeddings.pt'))
content_config.image = ImageContentConfig(
input_directories = ['tests/data/images'],
embeddings_file = model_dir.joinpath('image_embeddings.pt'),
batch_size = 1,
use_xmp_metadata = False)
return content_config

26
tests/test_org_to_jsonl.py
View File

@@ -15,7 +15,7 @@ def test_entry_with_empty_body_line_to_jsonl(tmp_path):
:PROPERTIES:
:ID: 42-42-42
:END:
\t\r\n
\t\r
'''
orgfile = create_file(tmp_path, entry)
@@ -37,7 +37,29 @@ def test_entry_with_body_to_jsonl(tmp_path):
:PROPERTIES:
:ID: 42-42-42
:END:
\t\r\nBody Line 1\n
\t\r
Body Line 1
'''
orgfile = create_file(tmp_path, entry)
# Act
# Extract Entries from specified Org files
entries, entry_to_file_map = extract_org_entries(org_files=[orgfile])
# Process Each Entry from All Notes Files
jsonl_string = convert_org_entries_to_jsonl(entries, entry_to_file_map)
jsonl_data = [json.loads(json_string) for json_string in jsonl_string.splitlines()]
# Assert
assert len(jsonl_data) == 1
def test_file_with_no_headings_to_jsonl(tmp_path):
"Ensure files with no heading, only body text are loaded."
# Arrange
entry = f'''
- Bullet point 1
- Bullet point 2
'''
orgfile = create_file(tmp_path, entry)

102
tests/test_orgnode.py
View File

@@ -8,6 +8,28 @@ from src.processor.org_mode import orgnode
# Test
# ----------------------------------------------------------------------------------------------------
def test_parse_entry_with_no_headings(tmp_path):
"Test parsing of entry with minimal fields"
# Arrange
entry = f'''Body Line 1'''
orgfile = create_file(tmp_path, entry)
# Act
entries = orgnode.makelist(orgfile)
# Assert
assert len(entries) == 1
assert entries[0].Heading() == f'{orgfile}'
assert entries[0].Tags() == set()
assert entries[0].Body() == "Body Line 1"
assert entries[0].Priority() == ""
assert entries[0].Property("ID") == ""
assert entries[0].Closed() == ""
assert entries[0].Scheduled() == ""
assert entries[0].Deadline() == ""
# ----------------------------------------------------------------------------------------------------
def test_parse_minimal_entry(tmp_path):
"Test parsing of entry with minimal fields"
@@ -81,18 +103,17 @@ Body Line 1
Body Line 2
'''
orgfile = create_file(tmp_path, entry)
normalized_orgfile = f'~/{relpath(orgfile, start=Path.home())}'
# Act
entries = orgnode.makelist(orgfile)
# Assert
# SOURCE link rendered with Heading
assert f':SOURCE: [[file:{normalized_orgfile}::*{entries[0].Heading()}]]' in f'{entries[0]}'
assert f':SOURCE: [[file:{orgfile}::*{entries[0].Heading()}]]' in f'{entries[0]}'
# ID link rendered with ID
assert f':ID: id:123-456-789-4234-1231' in f'{entries[0]}'
# LINE link rendered with line number
assert f':LINE: file:{normalized_orgfile}::2' in f'{entries[0]}'
assert f':LINE: file:{orgfile}::2' in f'{entries[0]}'
# ----------------------------------------------------------------------------------------------------
@@ -115,8 +136,7 @@ Body Line 1'''
# parsed heading from entry
assert entries[0].Heading() == "Heading[1]"
# ensure SOURCE link has square brackets in filename, heading escaped in rendered entries
normalized_orgfile = f'~/{relpath(orgfile, start=Path.home())}'
escaped_orgfile = f'{normalized_orgfile}'.replace("[1]", "\\[1\\]")
escaped_orgfile = f'{orgfile}'.replace("[1]", "\\[1\\]")
assert f':SOURCE: [[file:{escaped_orgfile}::*Heading\[1\]' in f'{entries[0]}'
@@ -168,10 +188,80 @@ Body 2
assert entry.Logbook() == [(datetime.datetime(1984,4,index+1,9,0,0), datetime.datetime(1984,4,index+1,12,0,0))]
# ----------------------------------------------------------------------------------------------------
def test_parse_entry_with_empty_title(tmp_path):
"Test parsing of entry with minimal fields"
# Arrange
entry = f'''#+TITLE:
Body Line 1'''
orgfile = create_file(tmp_path, entry)
# Act
entries = orgnode.makelist(orgfile)
# Assert
assert len(entries) == 1
assert entries[0].Heading() == f'{orgfile}'
assert entries[0].Tags() == set()
assert entries[0].Body() == "Body Line 1"
assert entries[0].Priority() == ""
assert entries[0].Property("ID") == ""
assert entries[0].Closed() == ""
assert entries[0].Scheduled() == ""
assert entries[0].Deadline() == ""
# ----------------------------------------------------------------------------------------------------
def test_parse_entry_with_title_and_no_headings(tmp_path):
"Test parsing of entry with minimal fields"
# Arrange
entry = f'''#+TITLE: test
Body Line 1'''
orgfile = create_file(tmp_path, entry)
# Act
entries = orgnode.makelist(orgfile)
# Assert
assert len(entries) == 1
assert entries[0].Heading() == 'test'
assert entries[0].Tags() == set()
assert entries[0].Body() == "Body Line 1"
assert entries[0].Priority() == ""
assert entries[0].Property("ID") == ""
assert entries[0].Closed() == ""
assert entries[0].Scheduled() == ""
assert entries[0].Deadline() == ""
# ----------------------------------------------------------------------------------------------------
def test_parse_entry_with_multiple_titles_and_no_headings(tmp_path):
"Test parsing of entry with minimal fields"
# Arrange
entry = f'''#+TITLE: title1
Body Line 1
#+TITLE: title2 '''
orgfile = create_file(tmp_path, entry)
# Act
entries = orgnode.makelist(orgfile)
# Assert
assert len(entries) == 1
assert entries[0].Heading() == 'title1 title2'
assert entries[0].Tags() == set()
assert entries[0].Body() == "Body Line 1\n"
assert entries[0].Priority() == ""
assert entries[0].Property("ID") == ""
assert entries[0].Closed() == ""
assert entries[0].Scheduled() == ""
assert entries[0].Deadline() == ""
# Helper Functions
def create_file(tmp_path, entry, filename="test.org"):
org_file = tmp_path / f"notes/{filename}"
org_file.parent.mkdir()
org_file.touch()
org_file.write_text(entry)
return org_file
return org_file

25
tests/test_text_search.py
View File

@@ -1,6 +1,10 @@
# System Packages
from copy import deepcopy
from pathlib import Path
# External Packages
import pytest
# Internal Packages
from src.utils.state import model
from src.search_type import text_search
@@ -9,6 +13,25 @@ from src.processor.org_mode.org_to_jsonl import org_to_jsonl
# Test
# ----------------------------------------------------------------------------------------------------
def test_asymmetric_setup_with_empty_file_raises_error(content_config: ContentConfig, search_config: SearchConfig):
# Arrange
file_to_index = Path(content_config.org.input_filter).parent / "new_file_to_index.org"
file_to_index.touch()
new_org_content_config = deepcopy(content_config.org)
new_org_content_config.input_files = [f'{file_to_index}']
new_org_content_config.input_filter = None
# Act
# Generate notes embeddings during asymmetric setup
with pytest.raises(ValueError, match=r'^No valid entries found*'):
text_search.setup(org_to_jsonl, new_org_content_config, search_config.asymmetric, regenerate=True)
# Cleanup
# delete created test file
file_to_index.unlink()
# ----------------------------------------------------------------------------------------------------
def test_asymmetric_setup(content_config: ContentConfig, search_config: SearchConfig):
# Act
@@ -23,7 +46,7 @@ def test_asymmetric_setup(content_config: ContentConfig, search_config: SearchCo
# ----------------------------------------------------------------------------------------------------
def test_asymmetric_search(content_config: ContentConfig, search_config: SearchConfig):
# Arrange
model.notes_search = text_search.setup(org_to_jsonl, content_config.org, search_config.asymmetric, regenerate=False)
model.notes_search = text_search.setup(org_to_jsonl, content_config.org, search_config.asymmetric, regenerate=True)
query = "How to git install application?"
# Act