string_tokenizer.c

00001 // Copyright (c) Microsoft. All rights reserved.
00002 // Licensed under the MIT license. See LICENSE file in the project root for full license information.
00003 
00004 #include <stdlib.h>
00005 #include "azure_c_shared_utility/gballoc.h"
00006 #include <stdbool.h>
00007 #include "azure_c_shared_utility/string_tokenizer.h"
00008 #include "azure_c_shared_utility/optimize_size.h"
00009 #include "azure_c_shared_utility/xlogging.h"
00010 #include "azure_c_shared_utility/crt_abstractions.h"
00011 
00012 typedef struct STRING_TOKEN_TAG
00013 {
00014     const char* inputString;
00015     const char* currentPos;
00016     size_t sizeOfinputString;
00017 } STRING_TOKEN;
00018 
00019 STRING_TOKENIZER_HANDLE STRING_TOKENIZER_create(STRING_HANDLE handle)
00020 {
00021     STRING_TOKENIZER_HANDLE result;
00022 
00023     /* Codes_SRS_STRING_04_001: [STRING_TOKENIZER_create shall return an NULL STRING_TOKENIZER_HANDLE if parameter handle is NULL] */
00024     if (handle == NULL)
00025     {
00026         LogError("Invalid Argument. Handle cannot be NULL.");
00027         result = NULL;
00028     }
00029     else
00030     {
00031         /* Codes_SRS_STRING_04_002: [STRING_TOKENIZER_create shall allocate a new STRING_TOKENIZER_HANDLE having the content of the STRING_HANDLE copied and current position pointing at the beginning of the string] */
00032         result = STRING_TOKENIZER_create_from_char(STRING_c_str(handle));
00033     }
00034 
00035     return result;
00036 }
00037 
00038 extern STRING_TOKENIZER_HANDLE STRING_TOKENIZER_create_from_char(const char* input)
00039 {
00040     STRING_TOKEN *result;
00041     char* inputStringToMalloc;
00042 
00043     /* Codes_SRS_STRING_07_001: [STRING_TOKENIZER_create shall return an NULL STRING_TOKENIZER_HANDLE if parameter input is NULL] */
00044     if (input == NULL)
00045     {
00046         LogError("Invalid Argument. Handle cannot be NULL.");
00047         result = NULL;
00048     }
00049     /* Codes_SRS_STRING_07_002: [STRING_TOKENIZER_create shall allocate a new STRING_TOKENIZER_HANDLE having the content of the STRING_HANDLE copied and current position pointing at the beginning of the string] */
00050     else if ((result = (STRING_TOKEN*)malloc(sizeof(STRING_TOKEN))) == NULL)
00051     {
00052         LogError("Memory Allocation failed. Cannot allocate STRING_TOKENIZER.");
00053     }
00054     else if ((mallocAndStrcpy_s(&inputStringToMalloc, input)) != 0)
00055     {
00056         LogError("Memory Allocation Failed. Cannot allocate and copy string Content.");
00057         free(result);
00058         result = NULL;
00059     }
00060     else
00061     {
00062         result->inputString = inputStringToMalloc;
00063         result->currentPos = result->inputString; //Current Pos will point to the initial position of Token.
00064         result->sizeOfinputString = strlen(result->inputString); //Calculate Size of Current String
00065     }
00066     return (STRING_TOKENIZER_HANDLE)result;
00067 }
00068 
00069 int STRING_TOKENIZER_get_next_token(STRING_TOKENIZER_HANDLE tokenizer, STRING_HANDLE output, const char* delimiters)
00070 {
00071     int result;
00072     /* Codes_SRS_STRING_04_004: [STRING_TOKENIZER_get_next_token shall return a nonzero value if any of the 3 parameters is NULL] */
00073     if (tokenizer == NULL || output == NULL || delimiters == NULL)
00074     {
00075         result = __FAILURE__;
00076     }
00077     else 
00078     {
00079         STRING_TOKEN* token = (STRING_TOKEN*)tokenizer;
00080         /* Codes_SRS_STRING_04_011: [Each subsequent call to STRING_TOKENIZER_get_next_token starts searching from the saved position on t and behaves as described above.] */
00081         size_t remainingInputStringSize = token->sizeOfinputString - (token->currentPos - token->inputString);
00082         size_t delimitterSize = strlen(delimiters);
00083 
00084         /* First Check if we reached the end of the string*/
00085         /* Codes_SRS_STRING_TOKENIZER_04_014: [STRING_TOKENIZER_get_next_token shall return nonzero value if t contains an empty string.] */
00086         if (remainingInputStringSize == 0)
00087         {
00088             result = __FAILURE__;
00089         }
00090         else if (delimitterSize == 0)
00091         {
00092             LogError("Empty delimiters parameter.");
00093             result = __FAILURE__;
00094         }
00095         else
00096         {
00097             size_t i;
00098             /* Codes_SRS_STRING_04_005: [STRING_TOKENIZER_get_next_token searches the string inside STRING_TOKENIZER_HANDLE for the first character that is NOT contained in the current delimiter] */
00099             for (i = 0; i < remainingInputStringSize; i++)
00100             {
00101                 size_t j;
00102 
00103                 bool foundDelimitter = false;
00104                 for (j = 0; j < delimitterSize; j++)
00105                 {
00106                     if (token->currentPos[i] == delimiters[j])
00107                     {
00108                         foundDelimitter = true;
00109                         break;
00110                     }
00111                 }
00112 
00113                 /* Codes_SRS_STRING_04_007: [If such a character is found, STRING_TOKENIZER_get_next_token consider it as the start of a token.] */
00114                 if (!foundDelimitter)
00115                 {
00116                     break;
00117                 }
00118             }
00119 
00120             /* Codes_SRS_STRING_04_006: [If no such character is found, then STRING_TOKENIZER_get_next_token shall return a nonzero Value (You've reach the end of the string or the string consists with only delimiters).] */
00121             //At this point update Current Pos to the character of the last token found or end of String.
00122             token->currentPos += i; 
00123             
00124             //Update the remainingInputStringSize
00125             remainingInputStringSize -= i;
00126             
00127             /* Codes_SRS_STRING_04_006: [If no such character is found, then STRING_TOKENIZER_get_next_token shall return a nonzero Value (You've reach the end of the string or the string consists with only delimiters).] */
00128             if (remainingInputStringSize == 0)
00129             {
00130                 result = __FAILURE__;
00131             }
00132             else
00133             {
00134                 bool foundDelimitter = false;
00135                 const char* endOfTokenPosition=NULL;
00136                 size_t amountOfCharactersToCopy;
00137                 size_t j;
00138                 //At this point the Current Pos is pointing to a character that is point to a nonDelimiter. So, now search for a Delimiter, till the end of the String. 
00139                 /*Codes_SRS_STRING_04_008: [STRING_TOKENIZER_get_next_token than searches from the start of a token for a character that is contained in the delimiters string.] */
00140                 /* Codes_SRS_STRING_04_009: [If no such character is found, STRING_TOKENIZER_get_next_token extends the current token to the end of the string inside t, copies the token to output and returns 0.] */
00141                 /* Codes_SRS_STRING_04_010: [If such a character is found, STRING_TOKENIZER_get_next_token consider it the end of the token and copy it's content to output, updates the current position inside t to the next character and returns 0.] */
00142                 for (j = 0; j < delimitterSize; j++)
00143                 {
00144                     if ((endOfTokenPosition = strchr(token->currentPos, delimiters[j])) != NULL)
00145                     {
00146                         foundDelimitter = true;
00147                         break;
00148                     }
00149                 }
00150 
00151                 //If token not found, than update the EndOfToken to the end of the inputString;
00152                 if (endOfTokenPosition == NULL)
00153                 {
00154                     amountOfCharactersToCopy = remainingInputStringSize;
00155                 }
00156                 else
00157                 {
00158                     amountOfCharactersToCopy = endOfTokenPosition - token->currentPos;
00159                 }
00160                 
00161                 //copy here the string to output. 
00162                 if (STRING_copy_n(output, token->currentPos, amountOfCharactersToCopy) != 0)
00163                 {
00164                     LogError("Problem copying token to output String.");
00165                     result = __FAILURE__;
00166                 }
00167                 else
00168                 {
00169                     //Update the Current position.
00170                     //Check if end of String reached so, currentPos points to the end of String.
00171                     if (foundDelimitter)
00172                     {
00173                         token->currentPos += amountOfCharactersToCopy + 1;
00174                     }
00175                     else
00176                     {
00177                         token->currentPos += amountOfCharactersToCopy;
00178                     }
00179                     
00180                     result = 0; //Result will be on the output. 
00181                 }
00182             }
00183         }
00184     }
00185 
00186     return result;
00187 }
00188 
00189 
00190 /* Codes_SRS_STRING_TOKENIZER_04_012: [STRING_TOKENIZER_destroy shall free the memory allocated by the STRING_TOKENIZER_create ] */
00191 void STRING_TOKENIZER_destroy(STRING_TOKENIZER_HANDLE t)
00192 {
00193     /* Codes_SRS_STRING_TOKENIZER_04_013: [When the t argument is NULL, then STRING_TOKENIZER_destroy shall not attempt to free] */
00194     if (t != NULL)
00195     {
00196         STRING_TOKEN* value = (STRING_TOKEN*)t;
00197         free((char*)value->inputString);
00198         value->inputString = NULL;
00199         free(value);
00200     }
00201 }
00202