|
| 1 | +//! 固定 Clang stdin SARIF 读者；不使用诊断自由文本或外部文件定位。 |
| 2 | +use serde_json::{Value, json}; |
| 3 | + |
| 4 | +/// 从固定Apple Clang21的SARIF提取同一冻结源码的原生规则及UTF-8字节位置。 |
| 5 | +/// 参数为有界报告、源码与原生成功状态；未知身份/列单位/报告矛盾返回错误。 |
| 6 | +pub fn parse_clang_stdin_sarif( |
| 7 | + bytes: &[u8], |
| 8 | + source: &[u8], |
| 9 | + success: bool, |
| 10 | +) -> Result<Vec<Value>, &'static str> { |
| 11 | + if bytes.len() > 64 * 1024 || source.len() > 1024 * 1024 { |
| 12 | + return Err("clang_report_invalid"); |
| 13 | + } |
| 14 | + let text = std::str::from_utf8(source).map_err(|_| "clang_report_invalid")?; |
| 15 | + let report = crate::parse_unique_json(bytes).map_err(|_| "clang_report_invalid")?; |
| 16 | + let runs = report["runs"] |
| 17 | + .as_array() |
| 18 | + .filter(|rows| rows.len() == 1) |
| 19 | + .ok_or("clang_report_invalid")?; |
| 20 | + let run = &runs[0]; |
| 21 | + let driver = &run["tool"]["driver"]; |
| 22 | + if report["version"] != "2.1.0" |
| 23 | + || run["columnKind"] != "unicodeCodePoints" |
| 24 | + || driver["name"] != "clang" |
| 25 | + || driver["version"] != "Apple clang version 21.0.0 (clang-2100.3.34.2)" |
| 26 | + || !run["invocations"] |
| 27 | + .as_array() |
| 28 | + .is_some_and(|rows| rows.len() == 1 && rows[0]["executionSuccessful"] == success) |
| 29 | + { |
| 30 | + return Err("clang_report_invalid"); |
| 31 | + } |
| 32 | + let results = run["results"] |
| 33 | + .as_array() |
| 34 | + .filter(|rows| rows.len() <= 64) |
| 35 | + .ok_or("clang_report_invalid")?; |
| 36 | + let rules = driver["rules"].as_array().ok_or("clang_report_invalid")?; |
| 37 | + let artifacts = run["artifacts"] |
| 38 | + .as_array() |
| 39 | + .filter(|rows| rows.len() <= 1) |
| 40 | + .ok_or("clang_report_invalid")?; |
| 41 | + if artifacts |
| 42 | + .iter() |
| 43 | + .any(|a| a["location"]["uri"] != "file://" || a["location"]["index"] != 0) |
| 44 | + { |
| 45 | + return Err("clang_report_invalid"); |
| 46 | + } |
| 47 | + let mut diagnostics = Vec::new(); |
| 48 | + let mut error_count = 0; |
| 49 | + for result in results { |
| 50 | + let id = result["ruleId"] |
| 51 | + .as_str() |
| 52 | + .filter(|s| { |
| 53 | + !s.is_empty() |
| 54 | + && s.len() <= 128 |
| 55 | + && s.bytes().all(|b| b.is_ascii_alphanumeric() || b == b'_') |
| 56 | + }) |
| 57 | + .ok_or("clang_report_invalid")?; |
| 58 | + let index = result["ruleIndex"] |
| 59 | + .as_u64() |
| 60 | + .and_then(|i| usize::try_from(i).ok()) |
| 61 | + .ok_or("clang_report_invalid")?; |
| 62 | + if rules.get(index).is_none_or(|r| r["id"] != id) { |
| 63 | + return Err("clang_report_invalid"); |
| 64 | + } |
| 65 | + let level = result["level"] |
| 66 | + .as_str() |
| 67 | + .filter(|level| matches!(*level, "error" | "warning" | "note")) |
| 68 | + .ok_or("clang_report_invalid")?; |
| 69 | + let locations = result["locations"] |
| 70 | + .as_array() |
| 71 | + .filter(|rows| !rows.is_empty() && rows.len() <= 4) |
| 72 | + .ok_or("clang_report_invalid")?; |
| 73 | + let mut first = None; |
| 74 | + for location in locations { |
| 75 | + let physical = &location["physicalLocation"]; |
| 76 | + if artifacts.len() != 1 |
| 77 | + || physical["artifactLocation"]["uri"] != "file://" |
| 78 | + || physical["artifactLocation"]["index"] != 0 |
| 79 | + { |
| 80 | + return Err("clang_report_invalid"); |
| 81 | + } |
| 82 | + let region = &physical["region"]; |
| 83 | + let line = region["startLine"] |
| 84 | + .as_u64() |
| 85 | + .filter(|n| *n > 0) |
| 86 | + .and_then(|n| usize::try_from(n).ok()) |
| 87 | + .ok_or("clang_report_invalid")?; |
| 88 | + let column = region["startColumn"] |
| 89 | + .as_u64() |
| 90 | + .filter(|n| *n > 0) |
| 91 | + .and_then(|n| usize::try_from(n).ok()) |
| 92 | + .ok_or("clang_report_invalid")?; |
| 93 | + let source_line = text |
| 94 | + .split('\n') |
| 95 | + .nth(line - 1) |
| 96 | + .ok_or("clang_report_invalid")?; |
| 97 | + let offset = if column == source_line.chars().count() + 1 { |
| 98 | + source_line.len() |
| 99 | + } else { |
| 100 | + source_line |
| 101 | + .char_indices() |
| 102 | + .nth(column - 1) |
| 103 | + .map(|(offset, _)| offset) |
| 104 | + .ok_or("clang_report_invalid")? |
| 105 | + }; |
| 106 | + if first.is_none() { |
| 107 | + first = Some( |
| 108 | + json!({"rule_id":format!("clang.{id}"),"line":line,"column_byte":offset+1,"level":level}), |
| 109 | + ); |
| 110 | + } |
| 111 | + } |
| 112 | + if level == "error" { |
| 113 | + error_count += 1; |
| 114 | + } |
| 115 | + if level != "note" { |
| 116 | + diagnostics.push(first.ok_or("clang_report_invalid")?); |
| 117 | + } |
| 118 | + } |
| 119 | + if success != (error_count == 0) { |
| 120 | + return Err("clang_report_invalid"); |
| 121 | + } |
| 122 | + Ok(diagnostics) |
| 123 | +} |
0 commit comments