#!/usr/bin/env python
import argparse
import json
import sarif_cli.traverse as S
import re
import sys
import collections

parser = argparse.ArgumentParser(description='summary of results')
parser.add_argument('file', metavar='sarif-file', type=str, help='input file, - for stdin')
parser.add_argument('-s', '--list-source', metavar='srcroot', type=str,
                    help='list source snippets using srcroot as sarif SRCROOT')
parser.add_argument('-r', '--related-locations', action="store_true",
                    help='list related locations like "hides [parameter](1)"')
parser.add_argument('-e', '--endpoints-only', action="store_true",
                    help='only list source and sink, dropping the path.  Identical, successive source/sink pairs are combined')
# TODO mutually exclusive options
parser.add_argument('-c', '--csv', action="store_true",
                    help='output csv instead of human-readable summary')

args = parser.parse_args()
with open(args.file, 'r') if args.file != '-' else sys.stdin as fp:
    sarif_struct = json.load(fp)

if args.csv:
    cw = S.get_csv_writer()

if not S.is_sarif_struct(sarif_struct):
    if args.csv:
        S.write_csv(cw, "ERROR", "invalid json contents %s", args.file)
    else:
        S.msg("ERROR: invalid json contents in %s\n" % (args.file))
    S.dbg("invalid json contents in %s\n" % (args.file))
    sys.exit(0)                 # No failure, just a warning

for runi in S.indices(sarif_struct, 'runs'):
    num_results = len(S.get(sarif_struct, 'runs', runi, 'results'))
    if num_results == 0: continue
    # 
    for resi in S.indices(sarif_struct, 'runs', runi, 'results'):
        result = S.get(sarif_struct, 'runs', runi, 'results', resi)
        if 'locations' in result:
            # Non-path problems
            # TODO: just pull out the uri, not the artifact
            message, artifact, region = S.get_location_message_info(result)
            if region == S.WholeFile:
                l1, c1, l2, c2 = -1, -1, -1, -1
            else:
                l1, c1, l2, c2 = S.lineinfo(region)
            filepath = "%s:%d:%d:%d:%d" % (artifact['uri'], l1, c1, l2, c2)
            if args.csv:
                S.write_csv(cw, "result", artifact['uri'], l1, c1, l2, c2, message)
            else:
                S.msg("RESULT: %s: %s\n" % (filepath, message))

            if region != S.WholeFile and args.list_source:
                lines = S.load_lines(args.list_source, artifact['uri'], l1, l2)
                if args.csv:
                    pass
                else:
                    for line, line_num in zip(lines, range(l1, l2+1)):
                        S.display_underlined(l1, c1, l2, c2, line, line_num)
            if args.related_locations:
                relatedLocations = result.get('relatedLocations', None)
                if type(relatedLocations) == list:
                    # Linking is explicit in output, so no need to get id(s) from message string.
                    for relo in relatedLocations:
                        message, artifact, region = S.get_relatedlocation_message_info(relo)
                        if artifact == S.NoFile:
                            if args.csv:
                                S.write_csv(cw, "result", "", -1, -1, -1, -1, message)
                            else:
                                S.msg("REFERENCE: %s: %s\n" % ("<NoFile>", message))
                        else:
                            if region == S.WholeFile:
                                l1, c1, l2, c2 = -1, -1, -1, -1
                            else:
                                l1, c1, l2, c2 = S.lineinfo(region)
                            filepath = "%s:%d:%d:%d:%d" % (artifact['uri'], l1, c1, l2, c2)
                            if args.csv:
                                S.write_csv(cw, "result", artifact['uri'], l1, c1, l2, c2, message)
                            else:
                                S.msg("REFERENCE: %s: %s\n" % (filepath, message))
                            if args.list_source:
                                lines = S.load_lines(args.list_source, artifact['uri'], l1, l2)
                                if args.csv:
                                    pass
                                else:
                                    for line, line_num in zip(lines, range(l1, l2+1)):
                                        S.display_underlined(l1, c1, l2, c2, line, line_num)
        if 'codeFlows' in result:
            # Path problems
            last_codeFlow = None
            for codefi in S.indices(result, 'codeFlows'):
                codeFlow = S.get(result, 'codeFlows', codefi)
                if args.csv:
                    S.write_csv(cw, "path", codefi)
                else:
                    S.msg("PATH %d\n" % codefi)
                    
                for threadi in S.indices(codeFlow, 'threadFlows'):
                    threadFlow = S.get(codeFlow, 'threadFlows', threadi)

                    if args.endpoints_only:
                        # 
                        # Pick the range to list only the endpoints (source/sink) of a threadFlow.
                        # 
                        t1 = S.indices(threadFlow, 'locations')
                        location_range = [t1[0], t1[-1]]
                        # 
                        # If the previous path had the same (source,sink) pair,
                        # we don't need to repeat it.
                        # 
                        if (last_codeFlow and
                            ( S.get(last_codeFlow, 'threadFlows', threadi, 'locations', 0) ==
                              S.get(codeFlow, 'threadFlows', threadi, 'locations', 0)) and
                            ( S.get(last_codeFlow, 'threadFlows', threadi, 'locations', -1) ==
                              S.get(codeFlow, 'threadFlows', threadi, 'locations', -1))):
                            continue
                    else:
                        location_range = S.indices(threadFlow, 'locations')

                    for loci in location_range:
                        location = S.get(threadFlow, 'locations', loci, 'location')
                        message, artifact, region = S.get_relatedlocation_message_info(location)
                        if artifact == S.NoFile:
                            if args.csv:
                                S.write_csv(cw, "flow_step", loci, "<NoFile>", -1, -1, -1, -1, message)
                            else:
                                S.msg("FLOW STEP %d: %s: %s\n" % (loci, "<NoFile>", message))
                        else:
                            if region == S.WholeFile:
                                l1, c1, l2, c2 = -1, -1, -1, -1
                            else:
                                l1, c1, l2, c2 = S.lineinfo(region)
                            filepath = "%s:%d:%d:%d:%d" % (artifact['uri'], l1, c1, l2, c2)
                            if args.csv:
                                S.write_csv(cw, "flow_step", loci, artifact['uri'], l1, c1, l2, c2, message)
                            else:
                                S.msg("FLOW STEP %d: %s: %s\n" % (loci, filepath, message))
                            if args.list_source:
                                lines = S.load_lines(args.list_source, artifact['uri'], l1, l2)
                                if args.csv:
                                    pass
                                else:
                                    for line, line_num in zip(lines, range(l1, l2+1)):
                                        S.display_underlined(l1, c1, l2, c2, line, line_num)
                last_codeFlow = codeFlow
        if args.csv:
            pass
        else:
            S.msg("\n")