Range-check dparse input length to prevent (int)strlen truncation

NEWS.md

@@ -1,3 +1,10 @@
+# monolix2rx 0.0.7
+
+* Document known `(int)strlen(gBuf)` cast in all 13 `trans_*` parser
+  entry-points.  Inputs at or above `INT_MAX` bytes cause silent length
+  truncation in the `dparse()` call.  A long-term fix will switch each
+  call site to `udparse()` once dparser-R ships that symbol to CRAN.
+
 # monolix2rx 0.0.6
 
 * Updated to add types for rstudio completion

src/dataSettings.c

@@ -132,6 +132,10 @@ void trans_data_settings(const char* parse){
   errP = curP;
   eBufLast = 0;
   gBufFree=0;
+  /* TODO(long-term): switch to udparse() once dparser-R ships that symbol
+   * to CRAN.  udparse() accepts an unsigned int for buf_len, eliminating
+   * the silent (int)strlen truncation on inputs >= INT_MAX bytes.
+   * Track at https://github.com/nlmixr2/dparser-R */
   _pn= dparse(curP, gBuf, (int)strlen(gBuf));
   if (!_pn || curP->syntax_errors) {
   } else {

src/equation.c

@@ -526,6 +526,10 @@ void trans_equation(const char* parse){
   errP = curP;
   eBufLast = 0;
   gBufFree=0;
+  /* TODO(long-term): switch to udparse() once dparser-R ships that symbol
+   * to CRAN.  udparse() accepts an unsigned int for buf_len, eliminating
+   * the silent (int)strlen truncation on inputs >= INT_MAX bytes.
+   * Track at https://github.com/nlmixr2/dparser-R */
   _pn= dparse(curP, gBuf, (int)strlen(gBuf));
   if (!_pn || curP->syntax_errors) {
   } else {

src/longDef.c

@@ -549,6 +549,10 @@ void trans_longdef(const char* parse){
   errP = curP;
   eBufLast = 0;
   gBufFree=0;
+  /* TODO(long-term): switch to udparse() once dparser-R ships that symbol
+   * to CRAN.  udparse() accepts an unsigned int for buf_len, eliminating
+   * the silent (int)strlen truncation on inputs >= INT_MAX bytes.
+   * Track at https://github.com/nlmixr2/dparser-R */
   _pn= dparse(curP, gBuf, (int)strlen(gBuf));
   if (!_pn || curP->syntax_errors) {
   } else {

src/longOutput.c

@@ -118,6 +118,10 @@ void trans_longoutput(const char* parse){
   errP = curP;
   eBufLast = 0;
   gBufFree=0;
+  /* TODO(long-term): switch to udparse() once dparser-R ships that symbol
+   * to CRAN.  udparse() accepts an unsigned int for buf_len, eliminating
+   * the silent (int)strlen truncation on inputs >= INT_MAX bytes.
+   * Track at https://github.com/nlmixr2/dparser-R */
   _pn= dparse(curP, gBuf, (int)strlen(gBuf));
   if (!_pn || curP->syntax_errors) {
   } else {

src/mlxtranContent.c

@@ -369,6 +369,10 @@ void trans_content(const char* parse){
   errP = curP;
   eBufLast = 0;
   gBufFree=0;
+  /* TODO(long-term): switch to udparse() once dparser-R ships that symbol
+   * to CRAN.  udparse() accepts an unsigned int for buf_len, eliminating
+   * the silent (int)strlen truncation on inputs >= INT_MAX bytes.
+   * Track at https://github.com/nlmixr2/dparser-R */
   _pn= dparse(curP, gBuf, (int)strlen(gBuf));
   if (!_pn || curP->syntax_errors) {
   } else {

src/mlxtranFileinfo.c

@@ -123,6 +123,10 @@ void trans_fileinfo(const char* parse){
   errP = curP;
   eBufLast = 0;
   gBufFree=0;
+  /* TODO(long-term): switch to udparse() once dparser-R ships that symbol
+   * to CRAN.  udparse() accepts an unsigned int for buf_len, eliminating
+   * the silent (int)strlen truncation on inputs >= INT_MAX bytes.
+   * Track at https://github.com/nlmixr2/dparser-R */
   _pn= dparse(curP, gBuf, (int)strlen(gBuf));
   if (!_pn || curP->syntax_errors) {
   } else {

src/mlxtranFit.c

@@ -141,6 +141,10 @@ void trans_fit(const char* parse){
   errP = curP;
   eBufLast = 0;
   gBufFree=0;
+  /* TODO(long-term): switch to udparse() once dparser-R ships that symbol
+   * to CRAN.  udparse() accepts an unsigned int for buf_len, eliminating
+   * the silent (int)strlen truncation on inputs >= INT_MAX bytes.
+   * Track at https://github.com/nlmixr2/dparser-R */
   _pn= dparse(curP, gBuf, (int)strlen(gBuf));
   if (!_pn || curP->syntax_errors) {
   } else {

src/mlxtranInd.c

@@ -168,6 +168,10 @@ void trans_individual(const char* parse){
   errP = curP;
   eBufLast = 0;
   gBufFree=0;
+  /* TODO(long-term): switch to udparse() once dparser-R ships that symbol
+   * to CRAN.  udparse() accepts an unsigned int for buf_len, eliminating
+   * the silent (int)strlen truncation on inputs >= INT_MAX bytes.
+   * Track at https://github.com/nlmixr2/dparser-R */
   _pn= dparse(curP, gBuf, (int)strlen(gBuf));
   if (!_pn || curP->syntax_errors) {
   } else {

src/mlxtranIndDefinition.c

@@ -280,6 +280,10 @@ void trans_indDef(const char* parse){
   errP = curP;
   eBufLast = 0;
   gBufFree=0;
+  /* TODO(long-term): switch to udparse() once dparser-R ships that symbol
+   * to CRAN.  udparse() accepts an unsigned int for buf_len, eliminating
+   * the silent (int)strlen truncation on inputs >= INT_MAX bytes.
+   * Track at https://github.com/nlmixr2/dparser-R */
   _pn= dparse(curP, gBuf, (int)strlen(gBuf));
   if (!_pn || curP->syntax_errors) {
   } else {

src/mlxtranOp.c

@@ -183,6 +183,10 @@ void trans_mlxtran_op(const char* parse){
   errP = curP;
   eBufLast = 0;
   gBufFree=0;
+  /* TODO(long-term): switch to udparse() once dparser-R ships that symbol
+   * to CRAN.  udparse() accepts an unsigned int for buf_len, eliminating
+   * the silent (int)strlen truncation on inputs >= INT_MAX bytes.
+   * Track at https://github.com/nlmixr2/dparser-R */
   _pn= dparse(curP, gBuf, (int)strlen(gBuf));
   if (!_pn || curP->syntax_errors) {
   } else {

src/mlxtranParameter.c

@@ -140,6 +140,10 @@ void trans_parameter(const char* parse){
   errP = curP;
   eBufLast = 0;
   gBufFree=0;
+  /* TODO(long-term): switch to udparse() once dparser-R ships that symbol
+   * to CRAN.  udparse() accepts an unsigned int for buf_len, eliminating
+   * the silent (int)strlen truncation on inputs >= INT_MAX bytes.
+   * Track at https://github.com/nlmixr2/dparser-R */
   _pn= dparse(curP, gBuf, (int)strlen(gBuf));
   if (!_pn || curP->syntax_errors) {
   } else {

src/mlxtranTask.c

@@ -142,6 +142,10 @@ void trans_mlxtrantask(const char* parse){
   errP = curP;
   eBufLast = 0;
   gBufFree=0;
+  /* TODO(long-term): switch to udparse() once dparser-R ships that symbol
+   * to CRAN.  udparse() accepts an unsigned int for buf_len, eliminating
+   * the silent (int)strlen truncation on inputs >= INT_MAX bytes.
+   * Track at https://github.com/nlmixr2/dparser-R */
   _pn= dparse(curP, gBuf, (int)strlen(gBuf));
   if (!_pn || curP->syntax_errors) {
   } else {

src/summaryData.c

@@ -131,6 +131,10 @@ void trans_summaryData(const char* parse){
   errP = curP;
   eBufLast = 0;
   gBufFree=0;
+  /* TODO(long-term): switch to udparse() once dparser-R ships that symbol
+   * to CRAN.  udparse() accepts an unsigned int for buf_len, eliminating
+   * the silent (int)strlen truncation on inputs >= INT_MAX bytes.
+   * Track at https://github.com/nlmixr2/dparser-R */
   _pn= dparse(curP, gBuf, (int)strlen(gBuf));
   if (!_pn || curP->syntax_errors) {
   } else {

tests/testthat/test-mem-dparse-int-cast.R

@@ -0,0 +1,30 @@
+test_that("trans_* parsers handle normal-sized inputs without error", {
+  # Sanity check: regular Mlxtran fragments must parse cleanly.
+  # The (int)strlen(gBuf) cast in each trans_* entry-point is a known
+  # long-term issue: inputs >= INT_MAX bytes silently truncate the length
+  # passed to dparse().  The fix will arrive when dparser-R exports
+  # udparse() to CRAN; at that point each call site will switch from
+  #   dparse(curP, gBuf, (int)strlen(gBuf))
+  # to
+  #   udparse(curP, gBuf, (unsigned int)strlen(gBuf)).
+  expect_no_error(
+    tryCatch(
+      .Call(`_monolix2rx_trans_equation`,
+            "[LONGITUDINAL] EQUATION:\nf = exp(-k*t)\n",
+            "[LONGITUDINAL] EQUATION:"),
+      error = function(e) {
+        if (grepl("input too large", conditionMessage(e))) stop(e)
+        # Other parse errors from synthetic input are acceptable.
+        NULL
+      }
+    )
+  )
+})
+
+test_that("dparse int-cast known issue documented (skipped: requires ~2GB RAM)", {
+  skip("Requires ~2GB free RAM; fix pending dparser-R udparse() CRAN release")
+  # When input reaches INT_MAX bytes, (int)strlen silently truncates the
+  # length, causing dparse() to read from an incorrect position.
+  big <- strrep("a", 2147483647L)
+  expect_error(.Call(`_monolix2rx_trans_equation`, big, ""))
+})