depyo 1.0.1 → 1.0.3

package/lib/PycReader.js

@@ -93,14 +93,29 @@ const MagicToVersion = {
     0x0A0D0D33: {major: 3, minor: 6, IsUnicode: true, opcode: require('./bytecode/python_3_6')},
     0x0A0D0D41: {major: 3, minor: 7, IsUnicode: true, opcode: require('./bytecode/python_3_7')},
     0x0A0D0D42: {major: 3, minor: 7, IsUnicode: true, opcode: require('./bytecode/python_3_7')},
-    0x0A0D0D49: {major: 3, minor: 8, IsUnicode: true, opcode: require('./bytecode/python_3_8')},
+    // Python 3.8a1 (3400) and 3.8a2 (3401) predate PEP 570 — no posonly field in marshal.
+    0x0A0D0D48: {major: 3, minor: 8, preReleaseMarshal: true, IsUnicode: true, opcode: require('./bytecode/python_3_8')},
+    0x0A0D0D49: {major: 3, minor: 8, preReleaseMarshal: true, IsUnicode: true, opcode: require('./bytecode/python_3_8')},
+    // 3.8a3 (3410) and later include posonly.
+    0x0A0D0D52: {major: 3, minor: 8, IsUnicode: true, opcode: require('./bytecode/python_3_8')},
+    0x0A0D0D53: {major: 3, minor: 8, IsUnicode: true, opcode: require('./bytecode/python_3_8')},
+    0x0A0D0D54: {major: 3, minor: 8, IsUnicode: true, opcode: require('./bytecode/python_3_8')},
     0x0A0D0D55: {major: 3, minor: 8, IsUnicode: true, opcode: require('./bytecode/python_3_8')},
     0x0A0D0D61: {major: 3, minor: 9, IsUnicode: true, opcode: require('./bytecode/python_3_9')},
     0x0A0D0D6F: {major: 3, minor: 10, IsUnicode: true, opcode: require('./bytecode/python_3_10')},
     0x0A0D0DA7: {major: 3, minor: 11, IsUnicode: true, opcode: require('./bytecode/python_3_11')},
     0x0A0D0DCB: {major: 3, minor: 12, IsUnicode: true, opcode: require('./bytecode/python_3_12')},
     0x0A0D0DF3: {major: 3, minor: 13, IsUnicode: true, opcode: require('./bytecode/python_3_13')},
-    0x0A0D0E2B: {major: 3, minor: 14, IsUnicode: true, opcode: require('./bytecode/python_3_14')}
+    0x0A0D0E2B: {major: 3, minor: 14, IsUnicode: true, opcode: require('./bytecode/python_3_14')},
+    0x0A0D0E4F: {major: 3, minor: 15, IsUnicode: true, opcode: require('./bytecode/python_3_15')},
+
+    // PyPy magics (shares opcode tables with matching CPython release).
+    // PyPy adds a handful of custom opcodes — unknowns degrade decode rather than fail.
+    0x0A0DF30A: {major: 2, minor: 7, IsUnicode: true,  opcode: require('./bytecode/python_2_7'), pypy: true},
+    0x0A0D0030: {major: 3, minor: 2, IsUnicode: true,  opcode: require('./bytecode/python_3_2'),  pypy: true},
+    0x0A0D0070: {major: 3, minor: 5, IsUnicode: true,  opcode: require('./bytecode/python_3_5'),  pypy: true},
+    0x0A0D00A0: {major: 3, minor: 6, IsUnicode: true,  opcode: require('./bytecode/python_3_6'),  pypy: true},
+    0x0A0D00F0: {major: 3, minor: 7, IsUnicode: true,  opcode: require('./bytecode/python_3_7'),  pypy: true}
 };
 
 const VersionAliases = {
@@ -109,6 +124,9 @@ const VersionAliases = {
 
 const VersionToInfo = {};
 for (const [magic, info] of Object.entries(MagicToVersion)) {
+    if (info.pypy) {
+        continue;
+    }
     const key = `${info.major}.${info.minor}`;
     const existing = VersionToInfo[key];
     const candidate = {...info, magic: Number(magic)};
@@ -280,6 +298,34 @@ class PycReader
         return best ? best.candidate : null;
     }
 
+    static ScanMarshalCandidates(buffer) {
+        const candidates = PycReader.ListSupportedVersions(false);
+        const results = [];
+        for (const candidate of candidates) {
+            const trial = PycReader.TryParseMarshal(buffer, candidate);
+            if (!trial) {
+                continue;
+            }
+            results.push({...trial, versionInfo: candidate});
+        }
+        results.sort((a, b) => {
+            if (a.unknownRatio !== b.unknownRatio) {
+                return a.unknownRatio - b.unknownRatio;
+            }
+            if (a.remaining !== b.remaining) {
+                return a.remaining - b.remaining;
+            }
+            if (a.unknown !== b.unknown) {
+                return a.unknown - b.unknown;
+            }
+            if (a.versionInfo.major !== b.versionInfo.major) {
+                return a.versionInfo.major - b.versionInfo.major;
+            }
+            return a.versionInfo.minor - b.versionInfo.minor;
+        });
+        return results;
+    }
+
     static CountUnknownOpcodes(codeObject, reader, opCodeList) {
         const code = codeObject?.Code?.Value;
         if (!code || !opCodeList) {
@@ -356,7 +402,7 @@ class PycReader
             if (obj.Names?.ClassName === "Py_Tuple") {
                 score += 1;
             }
-            return {score, remaining, unknown, unknownRatio};
+            return {score, remaining, unknown, total, unknownRatio};
         } catch (ex) {
             return null;
         }
@@ -598,7 +644,8 @@ class PycReader
         }
         codeObject.ArgCount = argCount;
 
-        codeObject.PosOnlyArgCount = this.versionCompare(3, 8) >= 0 ? this.m_rdr.readUInt32() : 0;
+        const hasPosOnly = this.versionCompare(3, 8) >= 0 && !this.m_version?.preReleaseMarshal;
+        codeObject.PosOnlyArgCount = hasPosOnly ? this.m_rdr.readUInt32() : 0;
         codeObject.KWOnlyArgCount = this.versionCompare(3, 0) >= 0 ? this.m_rdr.readUInt32() : 0;
         
         codeObject.NumLocals = 0;

package/lib/PythonObject.js

@@ -48,6 +48,29 @@ class PythonObject {
 
     // TODO: Refactor to use Symbol.toPrimitive() and Object.valueOf()
 
+    toReprString() {
+        if (this.ClassName === "Py_String" || this.ClassName === "Py_Unicode") {
+            let raw = this.Value;
+            if (raw == null || raw.length === 0) return '""';
+            raw = raw.toString();
+            let escaped = raw
+                .replace(/\\/g, '\\\\')
+                .replace(/\n/g, '\\n')
+                .replace(/\r/g, '\\r')
+                .replace(/\t/g, '\\t');
+            let quote = '"';
+            if (escaped.includes('"')) {
+                if (!escaped.includes("'")) {
+                    quote = "'";
+                } else {
+                    escaped = escaped.replace(/"/g, '\\"');
+                }
+            }
+            return quote + escaped + quote;
+        }
+        return this.toString();
+    }
+
     toString()
     {
         switch(this.ClassName) {
@@ -74,8 +97,16 @@ class PythonObject {
             case "Py_Interned":
                 return this.Value !== null ? this.Value.toString() : "0";
 
-            case "Py_Float":
-                return this.Value !== null ? `${this.Value}${Number.isInteger(this.Value) ? ".0" : ""}` : "0.0";
+            case "Py_Float": {
+                if (this.Value === null) return "0.0";
+                let s = `${this.Value}`;
+                // Python rejects `1e+300.0`; only append `.0` when the
+                // printed form is pure digits (optional sign).
+                if (/^-?\d+$/.test(s)) {
+                    s += ".0";
+                }
+                return s;
+            }
 
             case "Py_VeryLong":
                 if (this.Value) {
@@ -123,10 +154,11 @@ class PythonObject {
                 let res = "(";
                 if (this.Value) {
                     for (let obj of this.Value) {
+                        const part = obj instanceof PythonObject ? obj.toReprString() : String(obj);
                         if (res != "(") {
-                            res += ", " + obj;
+                            res += ", " + part;
                         } else {
-                            res += obj;
+                            res += part;
                         }
                     }
                     res += ")";
@@ -144,8 +176,10 @@ class PythonObject {
                     for (let pair of this.Value) {
                         if (res != "(") {
                             res += ", ";
-                        } 
-                        res += pair.key + ": " + pair.value;
+                        }
+                        const kPart = pair.key instanceof PythonObject ? pair.key.toReprString() : String(pair.key);
+                        const vPart = pair.value instanceof PythonObject ? pair.value.toReprString() : String(pair.value);
+                        res += kPart + ": " + vPart;
                     }
                     res += ")";
                     return res;