_c++-parseLsof.cpp

#include <string.h>
#include "json.hpp" // https://github.com/nlohmann/json/tree/develop/src
using json = nlohmann::json;

json parseLsof(std::string str) {
	json rez = json::array();

	json headers{ json::array() };
		// [ {name:'BLAH', min:0, max:0} ] // min max are indexes (1 col padding exclusive) (pad can be on both sides if in middle, or if first entry one on right, or if last entry then one on left)
	bool hdone = false; // header_done
	std::string word{ "" }; // current_word

	std::string::size_type lnstix{ 0 }; // line_start_index

	std::string::size_type ix{ 0 }; // ix throughout the whole string. its `i`
	std::string::size_type lnix{ 0 }; // ix on line offset by line start ix

	const char CHAR_N{ '\n' };
	const char CHAR_B{ ' ' };

	bool loopstarted = false;

	json entry = json::object();
	std::string::size_type hix{ 0 }; // header_index_currently_filling

	std::string::size_type LAST_CHAR_N{ str.find_last_of(CHAR_N) }; // line_index
	if (LAST_CHAR_N == std::string::npos) {
		// no new line char
		return rez;
	} else {
		LAST_CHAR_N += 1;
	}

	for(char& c : str) {
		if (loopstarted) ix++;
		if (!loopstarted) loopstarted = true;
		if (ix == LAST_CHAR_N) break; // as the lsof output ends in "\n0\u000b\u0001". I'm not sure about the final 3 chars, but at least the final "\n" sounds right

		// debug_log(ix, c);

		if (!hdone) {
			// space or \n means end of name
			if (c == CHAR_N || c == CHAR_B) {
				if (word.length()) {

					std::string::size_type min;
					std::string::size_type max;

					if (headers.size() == 0) {
						min = 0;
					} else {
						min = ix - word.length();
					}

					max = ix - 1;

					headers.push_back({
						{"name", word},
						{"min", min},
						{"max", max}
					});
					word = "";
				}
			} else {
				word += c;
			}

			if (c == CHAR_N) {
				hdone = true;
				lnstix = ix + 1;
				headers[headers.size()-1]["max"] = ix - 1;
			}
		} else {
			lnix = ix - lnstix;
			// debug_log( headers[hix]["name"], static_cast<int>(headers[hix]["min"]) );
			if (c == CHAR_N) {
				// start of block-link29393
				if (word.length()) {
					std::string cname = (headers[hix]["name"]);
					trim(word);
					entry[cname] = word;
					word = "";
					if (hix < headers.size()) hix++;
				}
				// end block-link29393
				for (int a_hix=hix; a_hix<headers.size(); a_hix++) {
					// hix is definitely NEXT, because if word had length then it was incremented ELSE word was blank so no work field added to entry for this
					std::string cname = (headers[a_hix]["name"]);
					entry[cname] = "";
				}
				rez.push_back(entry);
				entry.clear();
				hix = 0;
				lnstix = ix + 1;
			} else {
				std::string::size_type cmin = static_cast<int>(headers[hix]["min"]);
				std::string::size_type cmax = static_cast<int>(headers[hix]["max"]);

				if (lnix >= cmin && lnix <= cmax) {
					word += c;
				} else if (lnix > cmax) {
					char end_char = (hix == headers.size() - 1) ? CHAR_N : CHAR_B;

					if (c != end_char) {
						word += c;
						headers[hix]["max"] = lnix;
					} else {
						// copy of block-link29393
						if (word.length()) {
							std::string cname = (headers[hix]["name"]);
							trim(word);
							entry[cname] = word;
							word = "";
							if (hix < headers.size()) hix++;
						}
						// end block-link29393
					}
				} else if (lnix < cmin) {
					if (c != CHAR_B) {
						word += c;
						headers[hix]["min"] = lnix;
					}
				}
			}
		}
	}

	debug_log("headers:", headers.dump());
	debug_log("rez:", rez.dump());

	return rez;
}

usage.cpp

	std::string outstr = "COMMAND PID  USER   FD   TYPE DEVICE  SIZE/OFF   NODE NAME\nbash    252 noida  cwd    DIR    1,2      1088 410828 /Users/noida ion/Desktop\nbash    252 noida  txt    REG    1,2    628736  11647 /bin/bash\nbash    252 noida  txt    REG    1,2    622896  11866 /usr/lib/dyld\nbash    252 noida  txt    REG    1,2 382100934 418209 /private/var/db/dyld/dyld_shared_cache_x86_64\nbash    252 noida    0u   CHR   16,0  0t401079    601 /dev/ttys000\nbash    252 noida    1u   CHR   16,0  0t401079    601 /dev/ttys000\nbash    252 noida    2u   CHR   16,0  0t401079    601 /dev/ttys000\nbash    252 noida  255u   CHR   16,0  0t401079    601 /dev/ttys000\n0\u000b\u0001";
	json entries = parseLsof(outstr);
	debug_log("entries:", entries.dump());