XX Create new structure FileLocations to handle imported file errors

Author: Thirumalai-Shaktivel

.gitignore

@@ -62,7 +62,7 @@ share/jupyter/kernels/fortran/kernel.json
 src/runtime/*.o.empty.c
 python_ast.py
 python_ast.h
-ser.txt
+input.txt
 integration_tests/py_*
 integration_tests/b1/*
 integration_tests/b2/*
@@ -72,6 +72,7 @@ integration_tests/b5/*
 integration_tests/b6/*
 integration_tests/_lpython-tmp-test-*
 inst/bin/*
+*.tmp
 *.tlog
 *.filters
 *.obj

src/bin/lpython.cpp

@@ -380,9 +380,15 @@ class ASRToLLVMVisitor : public ASR::BaseVisitor<ASRToLLVMVisitor>
     void debug_get_line_column(const uint32_t &loc_first,
             uint32_t &line, uint32_t &column) {
         LocationManager lm;
-        lm.in_filename = infile;
-        lm.init_simple(LFortran::read_file(infile));
-        lm.pos_to_linecol(lm.output_to_input_pos(loc_first, false), line, column);
+        LocationManager::FileLocations fl;
+        fl.in_filename = infile;
+        lm.files.push_back(fl);
+        std::string input = LFortran::read_file(infile);
+// std::cout << infile << "\n" << input << "\n";
+        lm.init_simple(input);
+        lm.file_ends.push_back(input.size());
+        lm.pos_to_linecol(lm.output_to_input_pos(loc_first, false),
+            line, column, fl.in_filename);
     }
 
     template <typename T>

src/libasr/codegen/asr_to_llvm.cpp

@@ -61,20 +61,19 @@ bool Diagnostics::has_error() const {
     return false;
 }
 
-std::string Diagnostics::render(const std::string &input,
-        const LocationManager &lm, const CompilerOptions &compiler_options) {
+std::string Diagnostics::render(LocationManager &lm,
+        const CompilerOptions &compiler_options) {
     std::string out;
     for (auto &d : this->diagnostics) {
         if (compiler_options.no_warnings && d.level != Level::Error) {
             continue;
         }
         if (compiler_options.error_format == "human") {
-            out += render_diagnostic_human(d, input, lm,
-                compiler_options.use_colors,
+            out += render_diagnostic_human(d, lm, compiler_options.use_colors,
                 compiler_options.show_stacktrace);
             if (&d != &this->diagnostics.back()) out += "\n";
         } else if (compiler_options.error_format == "short") {
-            out += render_diagnostic_short(d, input, lm);
+            out += render_diagnostic_short(d, lm);
         } else {
             throw LCompilersException("Error format not supported.");
         }
@@ -108,49 +107,48 @@ std::string get_line(std::string str, int n)
     return line;
 }
 
-void populate_span(diag::Span &s, const LocationManager &lm,
-        const std::string &input) {
+void populate_span(diag::Span &s, const LocationManager &lm) {
     lm.pos_to_linecol(lm.output_to_input_pos(s.loc.first, false),
-        s.first_line, s.first_column);
+        s.first_line, s.first_column, s.filename);
     lm.pos_to_linecol(lm.output_to_input_pos(s.loc.last, true),
-        s.last_line, s.last_column);
-    s.filename = lm.in_filename;
+        s.last_line, s.last_column, s.filename);
+    std::string input;
+    read_file(s.filename, input);
+// std::cout << s.filename << "\n";
     for (uint32_t i = s.first_line; i <= s.last_line; i++) {
         s.source_code.push_back(get_line(input, i));
     }
     LFORTRAN_ASSERT(s.source_code.size() > 0)
 }
 
 // Loop over all labels and their spans, populate all of them
-void populate_spans(diag::Diagnostic &d, const LocationManager &lm,
-        const std::string &input) {
+void populate_spans(diag::Diagnostic &d, const LocationManager &lm) {
     for (auto &l : d.labels) {
         for (auto &s : l.spans) {
-            populate_span(s, lm, input);
+            populate_span(s, lm);
         }
     }
 }
 
 // Fills Diagnostic with span details and renders it
-std::string render_diagnostic_human(Diagnostic &d, const std::string &input,
-        const LocationManager &lm, bool use_colors, bool show_stacktrace) {
+std::string render_diagnostic_human(Diagnostic &d, const LocationManager &lm,
+        bool use_colors, bool show_stacktrace) {
     std::string out;
     if (show_stacktrace) {
         out += error_stacktrace(d.stacktrace);
     }
     // Convert to line numbers and get source code strings
-    populate_spans(d, lm, input);
+    populate_spans(d, lm);
     // Render the message
     out += render_diagnostic_human(d, use_colors);
     return out;
 }
 
 // Fills Diagnostic with span details and renders it
-std::string render_diagnostic_short(Diagnostic &d, const std::string &input,
-        const LocationManager &lm) {
+std::string render_diagnostic_short(Diagnostic &d, const LocationManager &lm) {
     std::string out;
     // Convert to line numbers and get source code strings
-    populate_spans(d, lm, input);
+    populate_spans(d, lm);
     // Render the message
     out += render_diagnostic_short(d);
     return out;
@@ -178,7 +176,7 @@ std::string render_diagnostic_human(const Diagnostic &d, bool use_colors) {
     std::string type_color = "";
     switch (d.level) {
         case (Level::Error):
-            primary_color = red_bold;    
+            primary_color = red_bold;
             type_color = primary_color;
             switch (d.stage) {
                 case (Stage::CPreprocessor):
@@ -205,17 +203,17 @@ std::string render_diagnostic_human(const Diagnostic &d, bool use_colors) {
             }
             break;
         case (Level::Warning):
-            primary_color = yellow_bold;    
+            primary_color = yellow_bold;
             type_color = primary_color;
             message_type = "warning";
             break;
         case (Level::Note):
-            primary_color = bold;    
+            primary_color = bold;
             type_color = primary_color;
             message_type = "note";
             break;
         case (Level::Help):
-            primary_color = bold;    
+            primary_color = bold;
             type_color = primary_color;
             message_type = "help";
             break;
@@ -286,6 +284,18 @@ std::string render_diagnostic_human(const Diagnostic &d, bool use_colors) {
                 }
                 // and start a new one:
                 s0 = s2;
+                if (s0.filename != s.filename) {
+                    out << std::endl;
+                    // TODO: print the primary line+column here, not the first label:
+                    out << std::string(line_num_width, ' ') << blue_bold;
+                    out << "-->" << reset << " " << s0.filename << ":";
+                    out << s0.first_line << ":" << s0.first_column;
+                    if (s0.first_line != s0.last_line) {
+                        out << " - " << s0.last_line << ":" << s0.last_column;
+                    }
+                    out << std::endl;
+                }
+
                 if (s0.first_line == s0.last_line) {
                     out << std::string(line_num_width+1, ' ') << blue_bold << "|"
                         << reset << std::endl;

src/libasr/diagnostics.cpp

@@ -113,8 +113,7 @@ struct Diagnostic {
 struct Diagnostics {
     std::vector<Diagnostic> diagnostics;
 
-    std::string render(const std::string &input,
-            const LocationManager &lm, const CompilerOptions &compiler_options);
+    std::string render(LocationManager &lm, const CompilerOptions &compiler_options);
 
     // Returns true iff diagnostics contains at least one error message
     bool has_error() const;
@@ -187,10 +186,9 @@ std::string render_diagnostic_human(const Diagnostic &d, bool use_colors);
 std::string render_diagnostic_short(const Diagnostic &d);
 
 // Fills Diagnostic with span details and renders it
-std::string render_diagnostic_human(Diagnostic &d, const std::string &input,
-        const LocationManager &lm, bool use_colors, bool show_stacktrace);
-std::string render_diagnostic_short(Diagnostic &d, const std::string &input,
-        const LocationManager &lm);
+std::string render_diagnostic_human(Diagnostic &d, const LocationManager &lm,
+    bool use_colors, bool show_stacktrace);
+std::string render_diagnostic_short(Diagnostic &d, const LocationManager &lm);
 
 } // namespace diag
 } // namespace LFortran

src/libasr/diagnostics.h

@@ -3,6 +3,7 @@
 
 #include <cstdint>
 #include <vector>
+// #include <iostream>
 
 namespace LFortran
 {
@@ -94,50 +95,67 @@ struct LocationManager {
     //
     //
     //
-    std::vector<uint32_t> out_start; // consecutive intervals in the output code
-    std::vector<uint32_t> in_start; // start + size in the original code
-    std::vector<uint32_t> in_newlines; // position of all \n in the original code
+    struct FileLocations {
+        std::vector<uint32_t> out_start; // consecutive intervals in the output code
+        std::vector<uint32_t> in_start; // start + size in the original code
+        std::vector<uint32_t> in_newlines; // position of all \n in the original code
 
-    // For preprocessor (if preprocessor==true).
-    // TODO: design a common structure, that also works with #include, that
-    // has these mappings for each file
-    bool preprocessor = false;
-    std::string in_filename;
-    uint32_t current_line=0;
-    std::vector<uint32_t> out_start0; // consecutive intervals in the output code
-    std::vector<uint32_t> in_start0; // start + size in the original code
-    std::vector<uint32_t> in_size0; // Size of the `in` interval
-    std::vector<uint32_t> interval_type0; // 0 .... 1:1; 1 ... many to many;
-    std::vector<uint32_t> in_newlines0; // position of all \n in the original code
-//    std::vector<uint32_t> filename_id; // file name for each interval, ID
-//    std::vector<std::string> filenames; // filenames lookup for an ID
+        // For preprocessor (if preprocessor==true).
+        // TODO: design a common structure, that also works with #include, that
+        // has these mappings for each file
+        bool preprocessor = false;
+        std::string in_filename;
+        uint32_t current_line=0;
+        std::vector<uint32_t> out_start0; // consecutive intervals in the output code
+        std::vector<uint32_t> in_start0; // start + size in the original code
+        std::vector<uint32_t> in_size0; // Size of the `in` interval
+        std::vector<uint32_t> interval_type0; // 0 .... 1:1; 1 ... many to many;
+        std::vector<uint32_t> in_newlines0; // position of all \n in the original code
+    };
+    std::vector<FileLocations> files;
+    std::vector<uint32_t> file_ends; // position of all ends of files
+    // For a given Location we use the `file_ends` and bisection to determine
+    // the file (index) which the location is from. Then we use this index into
+    // the `files` vector and use `in_newlines` and other information to
+    // determine the line and column inside this file, and the `in_filename`
+    // field to determine the filename. This happens when the diagnostic is
+    // printed. The `pos_to_linecol` function below should be modified to
+    // return line, column and the filename:
+    //
+    // void pos_to_linecol(uint32_t position, uint32_t &line, uint32_t &col,
+    //     std::string &filename) const;
 
     // Converts a position in the output code to a position in the original code
     // Every character in the output code has a corresponding location in the
     // original code, so this function always succeeds
     uint32_t output_to_input_pos(uint32_t out_pos, bool show_last) const {
-        if (out_start.size() == 0) return 0;
-        uint32_t interval = bisection(out_start, out_pos)-1;
-        uint32_t rel_pos = out_pos - out_start[interval];
-        uint32_t in_pos = in_start[interval] + rel_pos;
-        if (preprocessor) {
+// std::cout << "out: " << out_pos << "\n";
+        uint32_t index = bisection(file_ends, out_pos);
+        if (index == file_ends.size()) index -= 1;
+// std::cout << index << ":\n";
+        if (files[index].out_start.size() == 0) return 0;
+        uint32_t interval = bisection(files[index].out_start, out_pos)-1;
+        uint32_t rel_pos = out_pos - files[index].out_start[interval];
+        uint32_t in_pos = files[index].in_start[interval] + rel_pos;
+// std::cout <<  "in: "<< in_pos << "\n";
+        if (files[index].preprocessor) {
             // If preprocessor was used, do one more remapping
-            uint32_t interval0 = bisection(out_start0, in_pos)-1;
-            if (interval_type0[interval0] == 0) {
+            uint32_t interval0 = bisection(files[index].out_start0, in_pos)-1;
+            if (files[index].interval_type0[interval0] == 0) {
                 // 1:1 interval
-                uint32_t rel_pos0 = in_pos - out_start0[interval0];
-                uint32_t in_pos0 = in_start0[interval0] + rel_pos0;
+                uint32_t rel_pos0 = in_pos - files[index].out_start0[interval0];
+                uint32_t in_pos0 = files[index].in_start0[interval0] + rel_pos0;
                 return in_pos0;
             } else {
                 // many to many interval
                 uint32_t in_pos0;
-                if (in_pos == out_start0[interval0+1]-1 || show_last) {
+                if (in_pos == files[index].out_start0[interval0+1]-1 || show_last) {
                     // The end of the interval in "out" code
                     // Return the end of the interval in "in" code
-                    in_pos0 = in_start0[interval0]+in_size0[interval0]-1;
+                    in_pos0 = files[index].in_start0[interval0]+files[index].in_size0[interval0]-1;
                 } else {
                     // Otherwise return the beginning of the interval in "in"
-                    in_pos0 = in_start0[interval0];
+                    in_pos0 = files[index].in_start0[interval0];
                 }
                 return in_pos0;
             }
@@ -150,12 +168,26 @@ struct LocationManager {
     // `position` starts from 0
     // `line` and `col` starts from 1
     // `in_newlines` starts from 0
-    void pos_to_linecol(uint32_t position, uint32_t &line, uint32_t &col) const {
+    void pos_to_linecol(uint32_t position, uint32_t &line, uint32_t &col,
+            std::string &filename) const {
+        uint32_t index = bisection(file_ends, position);
+        if (index == file_ends.size()) index -= 1;
+// std::cout << position << ": Pos1\n";
+        filename = files[index].in_filename;
+// std::cout << "----------------------------\n";
+// for (auto &x: file_ends) std::cout << x << "\n";
+// std::cout << "----------------------------\n";
+// std::cout << file_ends[index]  << ": "<< file_ends[index - 1] << ": " << position - file_ends[index - 1] << "\n";
+        if (index > 0) position -= file_ends[index - 1];
+// std::cout << index << "_\n";
+// std::cout << position << ": Pos2\n";
+// std::cout << filename << ": Name\n";
+// exit(0);
         const std::vector<uint32_t> *newlines;
-        if (preprocessor) {
-            newlines = &in_newlines0;
+        if (files[index].preprocessor) {
+            newlines = &files[index].in_newlines0;
         } else {
-            newlines = &in_newlines;
+            newlines = &files[index].in_newlines;
         }
         int32_t interval = bisection(*newlines, position);
         if (interval >= 1 && position == (*newlines)[interval-1]) {
@@ -179,9 +211,9 @@ struct LocationManager {
 
     void init_simple(const std::string &input) {
         uint32_t n = input.size();
-        out_start = {0, n};
-        in_start = {0, n};
-        get_newlines(input, in_newlines);
+        files.back().out_start = {0, n};
+        files.back().in_start = {0, n};
+        get_newlines(input, files.back().in_newlines);
     }
 
 };

src/libasr/location.h

@@ -11,6 +11,7 @@ namespace LFortran::LPython {
               uint32_t first_column;
               uint32_t last_line;
               uint32_t last_column;
+              std::string filename;
               uint32_t severity;
        };
        struct document_symbols {
@@ -19,6 +20,7 @@ namespace LFortran::LPython {
               uint32_t first_column;
               uint32_t last_line;
               uint32_t last_column;
+              std::string filename;
        };
 
 } // namespace LFortran::Python

src/libasr/lsp_interface.h

@@ -99,7 +99,9 @@ std::string read_file(const std::string &filename)
     std::vector<char> bytes(filesize);
     ifs.read(&bytes[0], filesize);
 
-    return std::string(&bytes[0], filesize);
+    std::string s(&bytes[0], filesize);
+    s.erase( std::remove(s.begin(), s.end(), '\r'), s.end() );
+    return s;
 }
 
 std::string parent_path(const std::string &path) {

src/libasr/string_utils.cpp

@@ -25,6 +25,7 @@ bool read_file(const std::string &filename, std::string &text)
     ifs.read(&bytes[0], filesize);
 
     text = std::string(&bytes[0], filesize);
+    text.erase( std::remove(text.begin(), text.end(), '\r'), text.end() );
     return true;
 }