def print_grep_matches(grep_matches, color=True):
"""
Print text associated to grep matches.
Arguments:
grep_matches - the list of Artifacts to print
in the form of grep matches
color - a boolean, if True, color is enabled
"""
for view_id, match in enumerate(grep_matches):
path = "/".join(
fs.get_filename_parts_wo_prefix(match[0], config["PATH_KB_DATA"]))
line_number = match[1]
matched_text = match[2]
if color:
path = BOLD + str(path) + RESET
line_number = BOLD + str(line_number) + RESET
result_line = "{path}:{line_number}:{matched_text}".format(
path=path, line_number=line_number, matched_text=matched_text)
print(result_line)
def grep(args: Dict[str, str], config: Dict[str, str]):
"""
Grep through the list of artifacts of the knowledge base of kb.
Arguments:
args: - a dictionary containing the following fields:
regex -> the regex to search for
case_insensitive -> a boolean, if true,
the search will be case insensitive
matches -> a boolean, if true, only the raw
matches will be shown
verbose -> a boolean, if true, a verbose
output is produced on screen
config: - a configuration dictionary containing at least
the following keys:
PATH_KB_DB - the database path of KB
PATH_KB_DATA - the data directory of KB
PATH_KB_HIST - the history menu path of KB
"""
initializer.init(config)
conn = db.create_connection(config["PATH_KB_DB"])
# Get all artifacts
rows = db.get_artifacts_by_filter(conn, title="")
# Get all the file paths related to the artifacts in the database
file_list = [Path(config["PATH_KB_DATA"], r.category, r.title)
for r in rows]
# Grep in the files
results = fs.grep_in_files(
file_list,
args["regex"],
args["case_insensitive"])
# Get the list of artifact tuples in the form (category,title)
artifact_names = [fs.get_filename_parts_wo_prefix(
res[0], config["PATH_KB_DATA"]) for res in results]
# If user specied --matches -> just show matching lines and exit
if args["matches"]:
printer.print_grep_matches(artifact_names)
sys.exit(0)
# Get the set of uniq artifacts
uniq_artifact_names = set(artifact_names)
# Get the number of matches (hits) for each path found
filecounts = get_hits_per_artifact_name(artifact_names)
grep_result = list()
for art in uniq_artifact_names:
artifact = db.get_artifacts_by_filter(
conn, category=art[0], title=art[1])[0]
if artifact:
no_of_hits = filecounts[art]
grep_result.append((artifact, no_of_hits))
# Sort by number of hits, the largest -> the first
grep_result.sort(key=lambda x: x[1], reverse=True)
grep_artifacts = [r[0] for r in grep_result]
grep_hits = [r[1] for r in grep_result]
# Write to history file
history.write(config["PATH_KB_HIST"], grep_artifacts)
color_mode = not args["no_color"]
if args["verbose"]:
printer.print_grep_result_verbose(
grep_artifacts, grep_hits, color_mode)
else:
printer.print_grep_result(grep_artifacts, grep_hits, color_mode)