|
| 1 | +#!/usr/bin/python3 |
| 2 | + |
| 3 | +import json |
| 4 | +import os |
| 5 | +import os.path |
| 6 | +import shlex |
| 7 | +import subprocess |
| 8 | +import sys |
| 9 | +import tempfile |
| 10 | + |
| 11 | +class Generator: |
| 12 | + def __init__ (self, language): |
| 13 | + self.language = language |
| 14 | + self.generateSinks = False |
| 15 | + self.generateSources = False |
| 16 | + self.generateSummaries = False |
| 17 | + self.dryRun = False |
| 18 | + |
| 19 | + |
| 20 | + def printHelp(self): |
| 21 | + print(f"""Usage: |
| 22 | +python3 GenerateFlowModel.py <library-database> <outputQll> [--with-sinks] [--with-sources] [--with-summaries] [--dry-run] |
| 23 | +
|
| 24 | +This generates summary, source and sink models for the code in the database. |
| 25 | +The files will be placed in `{self.language}/ql/lib/semmle/code/{self.language}/frameworks/<outputQll>` where |
| 26 | +outputQll is the name (and path) of the output QLL file. Usually, models are grouped by their |
| 27 | +respective frameworks. |
| 28 | +
|
| 29 | +Which models are generated is controlled by the flags: |
| 30 | + --with-sinks |
| 31 | + --with-sources |
| 32 | + --with-summaries |
| 33 | +If none of these flags are specified, all models are generated. |
| 34 | +
|
| 35 | + --dry-run: Only run the queries, but don't write to file. |
| 36 | +
|
| 37 | +Example invocations: |
| 38 | +$ python3 GenerateFlowModel.py /tmp/dbs/my_library_db "mylibrary/Framework.qll" |
| 39 | +$ python3 GenerateFlowModel.py /tmp/dbs/my_library_db "mylibrary/FrameworkSinks.qll" --with-sinks |
| 40 | +
|
| 41 | +Requirements: `codeql` should both appear on your path. |
| 42 | + """) |
| 43 | + |
| 44 | + |
| 45 | + def setenvironment(self, target, database): |
| 46 | + self.codeQlRoot = subprocess.check_output(["git", "rev-parse", "--show-toplevel"]).decode("utf-8").strip() |
| 47 | + if not target.endswith(".qll"): |
| 48 | + target += ".qll" |
| 49 | + self.filename = os.path.basename(target) |
| 50 | + self.shortname = self.filename[:-4] |
| 51 | + self.database = database |
| 52 | + self.generatedFrameworks = os.path.join( |
| 53 | + self.codeQlRoot, f"{self.language}/ql/lib/semmle/code/{self.language}/frameworks/") |
| 54 | + self.frameworkTarget = os.path.join(self.generatedFrameworks, target) |
| 55 | + |
| 56 | + self.workDir = tempfile.mkdtemp() |
| 57 | + os.makedirs(self.generatedFrameworks, exist_ok=True) |
| 58 | + |
| 59 | + |
| 60 | + @staticmethod |
| 61 | + def make(language): |
| 62 | + generator = Generator(language) |
| 63 | + if any(s == "--help" for s in sys.argv): |
| 64 | + generator.printHelp() |
| 65 | + sys.exit(0) |
| 66 | + |
| 67 | + if "--with-sinks" in sys.argv: |
| 68 | + sys.argv.remove("--with-sinks") |
| 69 | + generator.generateSinks = True |
| 70 | + |
| 71 | + if "--with-sources" in sys.argv: |
| 72 | + sys.argv.remove("--with-sources") |
| 73 | + generator.generateSources = True |
| 74 | + |
| 75 | + if "--with-summaries" in sys.argv: |
| 76 | + sys.argv.remove("--with-summaries") |
| 77 | + generator.generateSummaries = True |
| 78 | + |
| 79 | + if "--dry-run" in sys.argv: |
| 80 | + sys.argv.remove("--dry-run") |
| 81 | + generator.dryRun = True |
| 82 | + |
| 83 | + if not generator.generateSinks and not generator.generateSources and not generator.generateSummaries: |
| 84 | + generator.generateSinks = generator.generateSources = generator.generateSummaries = True |
| 85 | + |
| 86 | + if len(sys.argv) != 3: |
| 87 | + generator.printHelp() |
| 88 | + sys.exit(1) |
| 89 | + |
| 90 | + generator.setenvironment(sys.argv[2], sys.argv[1]) |
| 91 | + return generator |
| 92 | + |
| 93 | + |
| 94 | + def runQuery(self, infoMessage, query): |
| 95 | + print("########## Querying " + infoMessage + "...") |
| 96 | + queryFile = os.path.join(self.codeQlRoot, f"{self.language}/ql/src/utils/model-generator", query) |
| 97 | + resultBqrs = os.path.join(self.workDir, "out.bqrs") |
| 98 | + cmd = ['codeql', 'query', 'run', queryFile, '--database', |
| 99 | + self.database, '--output', resultBqrs, '--threads', '8'] |
| 100 | + |
| 101 | + ret = subprocess.call(cmd) |
| 102 | + if ret != 0: |
| 103 | + print("Failed to generate " + infoMessage + |
| 104 | + ". Failed command was: " + shlex.join(cmd)) |
| 105 | + sys.exit(1) |
| 106 | + return self.readRows(resultBqrs) |
| 107 | + |
| 108 | + |
| 109 | + def readRows(self, bqrsFile): |
| 110 | + generatedJson = os.path.join(self.workDir, "out.json") |
| 111 | + cmd = ['codeql', 'bqrs', 'decode', bqrsFile, |
| 112 | + '--format=json', '--output', generatedJson] |
| 113 | + ret = subprocess.call(cmd) |
| 114 | + if ret != 0: |
| 115 | + print("Failed to decode BQRS. Failed command was: " + shlex.join(cmd)) |
| 116 | + sys.exit(1) |
| 117 | + |
| 118 | + with open(generatedJson) as f: |
| 119 | + results = json.load(f) |
| 120 | + |
| 121 | + try: |
| 122 | + results['#select']['tuples'] |
| 123 | + except KeyError: |
| 124 | + print('Unexpected JSON output - no tuples found') |
| 125 | + exit(1) |
| 126 | + |
| 127 | + rows = "" |
| 128 | + for (row) in results['#select']['tuples']: |
| 129 | + rows += " \"" + row[0] + "\",\n" |
| 130 | + |
| 131 | + return rows[:-2] |
| 132 | + |
| 133 | + |
| 134 | + def asCsvModel(self, superclass, kind, rows): |
| 135 | + classTemplate = """ |
| 136 | +private class {0}{1}Csv extends {2} {{ |
| 137 | + override predicate row(string row) {{ |
| 138 | + row = |
| 139 | + [ |
| 140 | +{3} |
| 141 | + ] |
| 142 | + }} |
| 143 | +}} |
| 144 | + """ |
| 145 | + if rows.strip() == "": |
| 146 | + return "" |
| 147 | + return classTemplate.format(self.shortname[0].upper() + self.shortname[1:], kind.capitalize(), superclass, rows) |
| 148 | + |
| 149 | + |
| 150 | + def makeContent(self): |
| 151 | + if self.generateSummaries: |
| 152 | + summaryRows = self.runQuery("summary models", "CaptureSummaryModels.ql") |
| 153 | + summaryCsv = self.asCsvModel("SummaryModelCsv", "summary", summaryRows) |
| 154 | + else: |
| 155 | + summaryCsv = "" |
| 156 | + |
| 157 | + if self.generateSinks: |
| 158 | + sinkRows = self.runQuery("sink models", "CaptureSinkModels.ql") |
| 159 | + sinkCsv = self.asCsvModel("SinkModelCsv", "sinks", sinkRows) |
| 160 | + else: |
| 161 | + sinkCsv = "" |
| 162 | + |
| 163 | + if self.generateSources: |
| 164 | + sourceRows = self.runQuery("source models", "CaptureSourceModels.ql") |
| 165 | + sourceCsv = self.asCsvModel("SourceModelCsv", "sources", sourceRows) |
| 166 | + else: |
| 167 | + sourceCsv = "" |
| 168 | + |
| 169 | + return f""" |
| 170 | +/** Definitions of taint steps in the {self.shortname} framework */ |
| 171 | +
|
| 172 | +import {self.language} |
| 173 | +private import semmle.code.{self.language}.dataflow.ExternalFlow |
| 174 | +
|
| 175 | +{sinkCsv} |
| 176 | +{sourceCsv} |
| 177 | +{summaryCsv} |
| 178 | +
|
| 179 | + """ |
| 180 | + |
| 181 | + |
| 182 | + def save(self, content): |
| 183 | + with open(self.frameworkTarget, "w") as frameworkQll: |
| 184 | + frameworkQll.write(content) |
| 185 | + |
| 186 | + cmd = ['codeql', 'query', 'format', '--in-place', self.frameworkTarget] |
| 187 | + ret = subprocess.call(cmd) |
| 188 | + if ret != 0: |
| 189 | + print("Failed to format query. Failed command was: " + shlex.join(cmd)) |
| 190 | + sys.exit(1) |
| 191 | + |
| 192 | + print("") |
| 193 | + print("CSV model written to " + self.frameworkTarget) |
| 194 | + |
| 195 | + |
| 196 | + def run(self): |
| 197 | + content = self.makeContent() |
| 198 | + |
| 199 | + if self.dryRun: |
| 200 | + print("CSV Models generated, but not written to file.") |
| 201 | + sys.exit(0) |
| 202 | + |
| 203 | + self.save(content) |
0 commit comments