Added compressFileAdv and advanced options

Author: darkdragn

src/python-lz4.c

@@ -41,6 +41,8 @@
 
 #define MAX(a, b)               ((a) > (b) ? (a) : (b))
 
+static PyObject *inputError;
+
 typedef int (*compressor)(const char *source, char *dest, int isize);
 
 static inline void store_le32(char *c, uint32_t x) {
@@ -55,6 +57,16 @@ static inline uint32_t load_le32(const char *c) {
     return d[0] | (d[1] << 8) | (d[2] << 16) | (d[3] << 24);
 }
 
+static inline char* add_extension(char* input) {
+    char* output;
+
+    output = (char*)malloc(strlen(input)+4);
+    strcpy(output, input);
+    strcat(output, ".lz4");
+
+    return output;
+}
+
 static const int hdr_size = sizeof(uint32_t);
 
 static PyObject *compress_with(compressor compress, PyObject *self, PyObject *args) {
@@ -129,25 +141,86 @@ static PyObject *py_lz4_uncompress(PyObject *self, PyObject *args) {
     return result;
 }
 
-
 static PyObject *py_lz4_compressFileDefault(PyObject *self, PyObject *args) {
     char* input;
     char* output;
-    int compLevel;
+    int compLevel = 0;
 
     (void)self;
-    if (!PyArg_ParseTuple(args, "si", &input, &compLevel)) {
+    if (!PyArg_ParseTuple(args, "s|i", &input, &compLevel)) {
         return NULL;
     }
 
-    output = (char*)malloc(strlen(input)+4);
-    strcpy(output, input);
-    strcat(output, ".lz4");
+    output = add_extension(input);
 
     LZ4IO_compressFilename(input, output, compLevel);
     return Py_None;
 }
 
+static PyObject *py_lz4_compressFileAdv(PyObject *self, PyObject *args, \
+                                        PyObject *keywds) {
+    char* input;
+    char* output = NULL;
+    int compLevel = 0;
+    int overwrite = NULL;
+    int blockSizeID = NULL;
+    int blockMode = 1;
+    int blockCheck = NULL;
+    int streamCheck = NULL;
+    int verbosity = NULL;
+
+    static char *kwlist[] = {"input", "compLevel", "output", "overwrite", 
+                             "blockSizeID", "blockMode", "blockCheck", 
+                             "streamCheck", "verbosity", NULL};
+    (void)self;
+    if (!PyArg_ParseTupleAndKeywords(args, keywds, "si|siiiiii", kwlist,
+                                     &input, &compLevel, &output, &overwrite, 
+                                     &blockSizeID, &blockMode, &blockCheck, 
+                                     &streamCheck, &verbosity)) {
+        return NULL;
+    }
+    
+    if (!output) { output = add_extension(input); }
+    if (overwrite) { LZ4IO_setOverwrite(overwrite); }
+    if (blockSizeID) {
+        if ((3 < blockSizeID) && (blockSizeID < 8)) {
+            LZ4IO_setBlockSizeID(blockSizeID);
+        }
+        else {
+            PyErr_SetString(inputError, "Invalid input for blockSizeID");
+        }
+    }
+    if ((blockMode == 0) || (blockMode == 1)) {
+        if (blockMode == 0 ) {LZ4IO_setBlockMode(chainedBlocks);} 
+    }
+    else {
+        PyErr_SetString(inputError, "Invalid input for blockMode");
+        return NULL;
+    }
+    if ((blockCheck == 0) || (blockCheck == 1)) {
+        if (blockCheck == 1) {LZ4IO_setBlockChecksumMode(blockCheck);}
+    }
+    else {
+        PyErr_SetString(inputError, "Invalid input for blockCheck");
+        return NULL;
+    }
+    if ((streamCheck == 0) || (streamCheck == 1)) {
+        if (streamCheck == 0) {LZ4IO_setStreamChecksumMode(streamCheck);}
+    }
+    else {
+        PyErr_SetString(inputError, "Invalid input for streamCheck");
+        return NULL;
+    }
+    if ((-1 < verbosity) && (verbosity < 5)) {
+        LZ4IO_setNotificationLevel(verbosity);
+    }
+    else {
+        PyErr_SetString(inputError, "Invalid input for verbosity");
+    }
+    
+    LZ4IO_compressFilename(input, output, compLevel);
+    return Py_None;
+}
 
 static PyObject *py_lz4_decompressFileDefault(PyObject *self, PyObject *args) {
     char* input;
@@ -163,8 +236,6 @@ static PyObject *py_lz4_decompressFileDefault(PyObject *self, PyObject *args) {
     output = (char*)calloc(outLen, sizeof(char));
     strncpy(output, input, outLen);
 
-    printf("%s \n", output);
-
     LZ4IO_decompressFilename(input, output);
     return Py_None;
 }
@@ -178,9 +249,10 @@ static PyMethodDef Lz4Methods[] = {
     {"uncompress",  py_lz4_uncompress, METH_VARARGS, UNCOMPRESS_DOCSTRING},
     {"decompress",  py_lz4_uncompress, METH_VARARGS, UNCOMPRESS_DOCSTRING},
     {"dumps",  py_lz4_compress, METH_VARARGS, COMPRESS_DOCSTRING},
-    {"loads",  py_lz4_uncompress, METH_VARARGS, UNCOMPRESS_DOCSTRING},
-    {"compressFileDefault", py_lz4_compressFileDefault, METH_VARARGS, COMPRESS_FILE_DOCSTRING},
-    {"decompressFileDefault", py_lz4_decompressFileDefault, METH_VARARGS, DECOMPRESS_FILE_DOCSTRING},
+    {"loads", py_lz4_uncompress, METH_VARARGS, UNCOMPRESS_DOCSTRING},
+    {"compressFileAdv", (PyCFunction)py_lz4_compressFileAdv, METH_VARARGS | METH_KEYWORDS, COMPF_ADV_DOCSTRING},
+    {"compressFileDefault", py_lz4_compressFileDefault, METH_VARARGS, COMPF_DEFAULT_DOCSTRING},
+    {"decompressFileDefault", py_lz4_decompressFileDefault, METH_VARARGS, DECOMP_FILE_DOCSTRING},
     {NULL, NULL, 0, NULL}
 };
 
@@ -190,6 +262,7 @@ struct module_state {
     PyObject *error;
 };
 
+
 #if PY_MAJOR_VERSION >= 3
 #define GETSTATE(m) ((struct module_state*)PyModule_GetState(m))
 #else
@@ -248,7 +321,9 @@ void initlz4(void)
         Py_DECREF(module);
         INITERROR;
     }
-
+    inputError = PyErr_NewException("input.error", NULL, NULL);
+    Py_INCREF(inputError);
+    PyModule_AddObject(module, "error", inputError);
 #if PY_MAJOR_VERSION >= 3
     return module;
 #endif

src/python-lz4.h

@@ -33,6 +33,7 @@
 
 static PyObject *py_lz4_compress(PyObject *self, PyObject *args);
 static PyObject *py_lz4_uncompress(PyObject *self, PyObject *args);
+static PyObject *py_lz4_compressFileAdv(PyObject *self, PyObject *args, PyObject *keywds);
 static PyObject *py_lz4_compressFileDefault(PyObject *self, PyObject *args);
 static PyObject *py_lz4_decompressFileDefault(PyObject *self, PyObject *args);
 
@@ -41,9 +42,13 @@ PyMODINIT_FUNC initlz4(void);
 #define COMPRESS_DOCSTRING      "Compress string, returning the compressed data.\nRaises an exception if any error occurs."
 #define COMPRESSHC_DOCSTRING    COMPRESS_DOCSTRING "\n\nCompared to compress, this gives a better compression ratio, but is much slower."
 #define UNCOMPRESS_DOCSTRING    "Decompress string, returning the uncompressed data.\nRaises an exception if any error occurs."
-#define COMPRESS_FILE_DOCSTRING "Compress file, by default adds .lz4 extension to original filename."\
-                                "\nAccepts two arguments, inputFile and compression level."
-#define DECOMPRESS_FILE_DOCSTRING   "Decompresses file, removes the extension by default, preserves original."
+#define COMPRESS_FILE_DOCSTRING "Compress file, by default adds .lz4 extension to original filename."
+#define COMPF_DEFAULT_DOCSTRING COMPRESS_FILE_DOCSTRING "\nAccepts two positional arguments, inputFile and compression level."
+#define COMPF_ADV_DOCSTRING     COMPRESS_FILE_DOCSTRING "\nRequires the first two keyword arugments and accepts any number of the"\
+                                "\nfollowing: input, compLevel, output, overwrite, blockSizeID, blockCheck, streamCheck"\
+                                "\nValid values are as follows(def=default): input='string', compLevel=0(low, def)-9(High), output='string',"\
+                                "\noverwrite=0/1(def), blockSizeID=4-7(def), blockCheck=0(def)/1, streamCheck=0/1(def), verbosity=0(def)-4"
+#define DECOMP_FILE_DOCSTRING   "Decompresses file, removes the extension by default, preserves original."
 
 #if defined(_WIN32) && defined(_MSC_VER)
 # define inline __inline

tests/test.py

@@ -1,16 +1,43 @@
+import hashlib
 import lz4
+import os
+import shutil
 import sys
-
-
 import unittest
-import os
 
 class TestLZ4(unittest.TestCase):
 
     def test_random(self):
       DATA = os.urandom(128 * 1024)  # Read 128kb
       self.assertEqual(DATA, lz4.loads(lz4.dumps(DATA)))
-
+    
+    def test_file(self):
+      fileName = 'src/lz4.c'
+      os.mkdir('testHold')
+      testNames = []
+      origDigest = hashlib.md5()
+      
+      with open('src/lz4.c', 'rb') as lz4Orig:
+        origDigest.update(lz4Orig.read())
+      
+      for num in range(1, 6):
+        testNames.append('testHold/test.%d.lz4' % num)
+      
+      lz4.compressFileAdv(fileName, 9, output=testNames[0])
+      lz4.compressFileAdv(fileName, 9, output=testNames[1], blockSizeID=4)
+      lz4.compressFileAdv(fileName, 9, output=testNames[2], blockSizeID=7)
+      lz4.compressFileAdv(fileName, 9, output=testNames[3], blockCheck=1)
+      lz4.compressFileAdv(fileName, 9, output=testNames[4], streamCheck=0)
+      
+      for test in testNames:
+        lz4.decompressFileDefault(test)
+        testDigest = hashlib.md5()
+        with open(test.replace('.lz4', ''), 'rb') as testFile:
+          testDigest.update(testFile.read())
+        self.assertEqual(origDigest.hexdigest(), testDigest.hexdigest())
+        del testDigest
+      shutil.rmtree('testHold')
+      
 if __name__ == '__main__':
     unittest.main()