Avoid casts that don't change the pointer type (#6785)

* Don't cast LOGICAL() to int*

If the returned pointer type ever changes, the absence of the cast will
give us a compiler warning due to the value being implicitly cast to an
incompatible pointer type. An explicit cast, on the other hand, tells
the compiler to silence the warning and do as written.

* Drop more no-op casts

* Introduce the Coccinelle linter

For mistakes that are easy to fix automatically, Coccinelle patches may
be written. If a patch finds a mistake, the linter will ask the human to
fix either the code or the patch.

Author: aitap

.ci/linters/c/cocci_linter.R

@@ -0,0 +1,20 @@
+cocci_linter = if (!nzchar(Sys.which("spatch"))) function(...) {} else function(c_obj) {
+  bad <- FALSE
+  for (spfile in list.files(".ci/linters/cocci", full.names = TRUE)) {
+    # Coccinelle parser gets confused sometimes, so ignore stderr and the exit code
+    out = suppressWarnings(system2(
+      "spatch",
+      shQuote(c(
+        "--sp-file", spfile, c_obj$path, "--recursive-includes",
+        "-I", R.home("include"), "-I", "src"
+      )),
+      stdout = TRUE, stderr = FALSE
+    ))
+    if (length(out) > 0) {
+      cat(sprintf("In file '%s', Coccinelle patch '%s' recommends the following changes:\n", c_obj$path, spfile))
+      writeLines(out)
+      bad <- TRUE
+    }
+  }
+  if (bad) stop("Please apply the changes above or fix the linter")
+}

.ci/linters/cocci/redundant_cast_to_same_pointer_type.cocci

@@ -0,0 +1,6 @@
+@@
+type T;
+T* E;
+@@
+- (T*)
+  E

.github/workflows/code-quality.yaml

@@ -35,6 +35,9 @@ jobs:
     steps:
       - uses: actions/checkout@v4
       - uses: r-lib/actions/setup-r@v2
+      - name: Install Coccinelle
+        # relying on the action above us to have updated the package cache
+        run: /usr/bin/sudo apt-get -y install coccinelle
       - name: Lint
         run: |
           linter_env = new.env()

src/assign.c

@@ -1207,7 +1207,7 @@ void writeNA(SEXP v, const int from, const int n, const bool listNA)
     memset(RAW(v)+from, 0, n*sizeof(Rbyte));
     break;
   case LGLSXP: {
-    int *vd = (int *)LOGICAL(v);
+    int *vd = LOGICAL(v);
     for (int i=from; i<=to; ++i) vd[i] = NA_LOGICAL;
   } break;
   case INTSXP: {

src/coalesce.c

@@ -100,7 +100,7 @@ SEXP coalesce(SEXP x, SEXP inplaceArg) {
         if (val!=NA_INTEGER64) xP[i]=val; else if (final) xP[i]=finalVal;
       }
     } else {
-      double *xP = (double *)REAL(first), finalVal=NA_REAL;
+      double *xP = REAL(first), finalVal=NA_REAL;
       int k=0;
       for (int j=0; j<nval; ++j) {
         SEXP item = VECTOR_ELT(x, j+off);

src/fread.c

@@ -441,7 +441,7 @@ static const char* filesize_to_str(size_t fsize)
 double copyFile(size_t fileSize)  // only called in very very rare cases
 {
   double tt = wallclock();
-  mmp_copy = (char *)malloc((size_t)fileSize + 1 /* extra \0 */);
+  mmp_copy = (char *)malloc(fileSize + 1 /* extra \0 */);
   if (!mmp_copy)
     return -1.0; // # nocov
   memcpy(mmp_copy, mmp, fileSize);

src/freadR.c

@@ -624,7 +624,7 @@ void pushBuffer(ThreadLocalFreadParsingContext *ctx)
     resj++;
     if (type[j]!=CT_STRING && type[j]>0) {
       if (thisSize == 8) {
-        double *dest = (double *)REAL(VECTOR_ELT(DT, resj)) + DTi;
+        double *dest = REAL(VECTOR_ELT(DT, resj)) + DTi;
         const char *src8 = (char*)buff8 + off8;
         for (int i=0; i<nRows; ++i) {
           *dest = *(double *)src8;
@@ -633,7 +633,7 @@ void pushBuffer(ThreadLocalFreadParsingContext *ctx)
         }
       } else
       if (thisSize == 4) {
-        int *dest = (int *)INTEGER(VECTOR_ELT(DT, resj)) + DTi;
+        int *dest = INTEGER(VECTOR_ELT(DT, resj)) + DTi;
         const char *src4 = (char*)buff4 + off4;
         // debug line for #3369 ... if (DTi>2638000) printf("freadR.c:460: thisSize==4, resj=%d, %"PRIu64", %d, %d, j=%d, done=%d\n", resj, (uint64_t)DTi, off4, rowSize4, j, done);
         for (int i=0; i<nRows; ++i) {

src/ijoin.c

@@ -10,8 +10,8 @@ SEXP lookup(SEXP ux, SEXP xlen, SEXP indices, SEXP gaps, SEXP overlaps, SEXP mul
 
   SEXP vv, tt, lookup, type_lookup;
   R_len_t *idx,*count,*type_count,xrows=INTEGER(xlen)[0],uxrows=LENGTH(VECTOR_ELT(ux, 0)),uxcols=LENGTH(ux);
-  int *from = (int *)INTEGER(VECTOR_ELT(indices, 0));
-  int *to   = (int *)INTEGER(VECTOR_ELT(indices, 1));
+  int *from = INTEGER(VECTOR_ELT(indices, 0));
+  int *to   = INTEGER(VECTOR_ELT(indices, 1));
   clock_t pass1, pass2, pass3, start;
   enum {ALL, FIRST, LAST} mult = ALL;
   enum {ANY, WITHIN, START, END, EQUAL} type = ANY;
@@ -31,8 +31,8 @@ SEXP lookup(SEXP ux, SEXP xlen, SEXP indices, SEXP gaps, SEXP overlaps, SEXP mul
   // For reference: uxcols-1 = type_count, uxcols-2 = count, uxcols-3 = type_lookup, uxcols-4 = lookup
   // first pass: calculate lengths first
   start = clock();
-  count = (int *)INTEGER(VECTOR_ELT(ux, uxcols-2));
-  type_count = (int *)INTEGER(VECTOR_ELT(ux, uxcols-1));
+  count = INTEGER(VECTOR_ELT(ux, uxcols-2));
+  type_count = INTEGER(VECTOR_ELT(ux, uxcols-1));
   switch (mult) {
   case FIRST:
     switch(type) {
@@ -225,10 +225,10 @@ SEXP overlaps(SEXP ux, SEXP imatches, SEXP multArg, SEXP typeArg, SEXP nomatchAr
 
   R_len_t uxcols=LENGTH(ux),rows=length(VECTOR_ELT(imatches,0));
   int nomatch = INTEGER(nomatchArg)[0], totlen=0, thislen;
-  int *from   = (int *)INTEGER(VECTOR_ELT(imatches, 0));
-  int *to     = (int *)INTEGER(VECTOR_ELT(imatches, 1));
-  int *count   = (int *)INTEGER(VECTOR_ELT(ux, uxcols-2));
-  int *type_count   = (int *)INTEGER(VECTOR_ELT(ux, uxcols-1));
+  int *from   = INTEGER(VECTOR_ELT(imatches, 0));
+  int *to     = INTEGER(VECTOR_ELT(imatches, 1));
+  int *count   = INTEGER(VECTOR_ELT(ux, uxcols-2));
+  int *type_count   = INTEGER(VECTOR_ELT(ux, uxcols-1));
   SEXP lookup = VECTOR_ELT(ux, uxcols-4);
   SEXP type_lookup = VECTOR_ELT(ux, uxcols-3);
   SEXP ans, f1__, f2__, tmp1, tmp2;

src/negate.c

@@ -5,7 +5,7 @@ void negateByRef(SEXP x) {
     error("not logical or integer vector");  // # nocov
   }
   const int n = length(x);
-  int *ansd = (int *)LOGICAL(x);
+  int *ansd = LOGICAL(x);
   for(int i=0; i<n; ++i) {
     ansd[i] ^= (ansd[i] != NA_LOGICAL);  // invert true/false but leave NA alone
   }

src/reorder.c

@@ -63,7 +63,7 @@ SEXP reorder(SEXP x, SEXP order)
     // This check is once up front, and then idx is applied to all the columns which is where the most time is spent.
   }
 
-  char *TMP = (char *)R_alloc(nmid, maxSize);
+  char *TMP = R_alloc(nmid, maxSize);
 
   for (int i=0; i<ncol; ++i) {
     const SEXP v = isNewList(x) ? VECTOR_ELT(x,i) : x;