1 //===-- runtime/format-implementation.h -------------------------*- C++ -*-===//
2 //
3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4 // See https://llvm.org/LICENSE.txt for license information.
5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6 //
7 //===----------------------------------------------------------------------===//
8 
9 // Implements out-of-line member functions of template class FormatControl
10 
11 #ifndef FORTRAN_RUNTIME_FORMAT_IMPLEMENTATION_H_
12 #define FORTRAN_RUNTIME_FORMAT_IMPLEMENTATION_H_
13 
14 #include "format.h"
15 #include "io-stmt.h"
16 #include "main.h"
17 #include "flang/Common/format.h"
18 #include "flang/Decimal/decimal.h"
19 #include <algorithm>
20 #include <limits>
21 
22 namespace Fortran::runtime::io {
23 
24 template <typename CONTEXT>
25 FormatControl<CONTEXT>::FormatControl(const Terminator &terminator,
26     const CharType *format, std::size_t formatLength, int maxHeight)
27     : maxHeight_{static_cast<std::uint8_t>(maxHeight)}, format_{format},
28       formatLength_{static_cast<int>(formatLength)} {
29   RUNTIME_CHECK(terminator, maxHeight == maxHeight_);
30   RUNTIME_CHECK(
31       terminator, formatLength == static_cast<std::size_t>(formatLength_));
32   stack_[0].start = offset_;
33   stack_[0].remaining = Iteration::unlimited; // 13.4(8)
34 }
35 
36 template <typename CONTEXT>
37 int FormatControl<CONTEXT>::GetMaxParenthesisNesting(
38     IoErrorHandler &handler, const CharType *format, std::size_t formatLength) {
39   int maxNesting{0};
40   int nesting{0};
41   const CharType *end{format + formatLength};
42   std::optional<CharType> quote;
43   int repeat{0};
44   for (const CharType *p{format}; p < end; ++p) {
45     if (quote) {
46       if (*p == *quote) {
47         quote.reset();
48       }
49     } else if (*p >= '0' && *p <= '9') {
50       repeat = 10 * repeat + *p - '0';
51     } else if (*p != ' ') {
52       switch (*p) {
53       case '\'':
54       case '"':
55         quote = *p;
56         break;
57       case 'h':
58       case 'H': // 9HHOLLERITH
59         p += repeat;
60         if (p >= end) {
61           handler.SignalError(IostatErrorInFormat,
62               "Hollerith (%dH) too long in FORMAT", repeat);
63           return maxNesting;
64         }
65         break;
66       case ' ':
67         break;
68       case '(':
69         ++nesting;
70         maxNesting = std::max(nesting, maxNesting);
71         break;
72       case ')':
73         nesting = std::max(nesting - 1, 0);
74         break;
75       }
76       repeat = 0;
77     }
78   }
79   if (quote) {
80     handler.SignalError(
81         IostatErrorInFormat, "Unbalanced quotation marks in FORMAT string");
82   } else if (nesting) {
83     handler.SignalError(
84         IostatErrorInFormat, "Unbalanced parentheses in FORMAT string");
85   }
86   return maxNesting;
87 }
88 
89 template <typename CONTEXT>
90 int FormatControl<CONTEXT>::GetIntField(
91     IoErrorHandler &handler, CharType firstCh) {
92   CharType ch{firstCh ? firstCh : PeekNext()};
93   if (ch != '-' && ch != '+' && (ch < '0' || ch > '9')) {
94     handler.SignalError(IostatErrorInFormat,
95         "Invalid FORMAT: integer expected at '%c'", static_cast<char>(ch));
96     return 0;
97   }
98   int result{0};
99   bool negate{ch == '-'};
100   if (negate) {
101     firstCh = '\0';
102     ch = PeekNext();
103   }
104   while (ch >= '0' && ch <= '9') {
105     if (result >
106         std::numeric_limits<int>::max() / 10 - (static_cast<int>(ch) - '0')) {
107       handler.SignalError(
108           IostatErrorInFormat, "FORMAT integer field out of range");
109       return result;
110     }
111     result = 10 * result + ch - '0';
112     if (firstCh) {
113       firstCh = '\0';
114     } else {
115       ++offset_;
116     }
117     ch = PeekNext();
118   }
119   if (negate && (result *= -1) > 0) {
120     handler.SignalError(
121         IostatErrorInFormat, "FORMAT integer field out of range");
122   }
123   return result;
124 }
125 
126 template <typename CONTEXT>
127 static void HandleControl(CONTEXT &context, char ch, char next, int n) {
128   MutableModes &modes{context.mutableModes()};
129   switch (ch) {
130   case 'B':
131     if (next == 'Z') {
132       modes.editingFlags |= blankZero;
133       return;
134     }
135     if (next == 'N') {
136       modes.editingFlags &= ~blankZero;
137       return;
138     }
139     break;
140   case 'D':
141     if (next == 'C') {
142       modes.editingFlags |= decimalComma;
143       return;
144     }
145     if (next == 'P') {
146       modes.editingFlags &= ~decimalComma;
147       return;
148     }
149     break;
150   case 'P':
151     if (!next) {
152       modes.scale = n; // kP - decimal scaling by 10**k
153       return;
154     }
155     break;
156   case 'R':
157     switch (next) {
158     case 'N':
159       modes.round = decimal::RoundNearest;
160       return;
161     case 'Z':
162       modes.round = decimal::RoundToZero;
163       return;
164     case 'U':
165       modes.round = decimal::RoundUp;
166       return;
167     case 'D':
168       modes.round = decimal::RoundDown;
169       return;
170     case 'C':
171       modes.round = decimal::RoundCompatible;
172       return;
173     case 'P':
174       modes.round = executionEnvironment.defaultOutputRoundingMode;
175       return;
176     default:
177       break;
178     }
179     break;
180   case 'X':
181     if (!next) {
182       context.HandleRelativePosition(n);
183       return;
184     }
185     break;
186   case 'S':
187     if (next == 'P') {
188       modes.editingFlags |= signPlus;
189       return;
190     }
191     if (!next || next == 'S') {
192       modes.editingFlags &= ~signPlus;
193       return;
194     }
195     break;
196   case 'T': {
197     if (!next) { // Tn
198       context.HandleAbsolutePosition(n - 1); // convert 1-based to 0-based
199       return;
200     }
201     if (next == 'L' || next == 'R') { // TLn & TRn
202       context.HandleRelativePosition(next == 'L' ? -n : n);
203       return;
204     }
205   } break;
206   default:
207     break;
208   }
209   if (next) {
210     context.SignalError(IostatErrorInFormat,
211         "Unknown '%c%c' edit descriptor in FORMAT", ch, next);
212   } else {
213     context.SignalError(
214         IostatErrorInFormat, "Unknown '%c' edit descriptor in FORMAT", ch);
215   }
216 }
217 
218 // Locates the next data edit descriptor in the format.
219 // Handles all repetition counts and control edit descriptors.
220 // Generally assumes that the format string has survived the common
221 // format validator gauntlet.
222 template <typename CONTEXT>
223 int FormatControl<CONTEXT>::CueUpNextDataEdit(Context &context, bool stop) {
224   int unlimitedLoopCheck{-1};
225   while (true) {
226     std::optional<int> repeat;
227     bool unlimited{false};
228     CharType ch{GetNextChar(context)};
229     while (ch == ',' || ch == ':') {
230       // Skip commas, and don't complain if they're missing; the format
231       // validator does that.
232       if (stop && ch == ':') {
233         return 0;
234       }
235       ch = GetNextChar(context);
236     }
237     if (ch == '-' || ch == '+' || (ch >= '0' && ch <= '9')) {
238       repeat = GetIntField(context, ch);
239       ch = GetNextChar(context);
240     } else if (ch == '*') {
241       unlimited = true;
242       ch = GetNextChar(context);
243       if (ch != '(') {
244         context.SignalError(IostatErrorInFormat,
245             "Invalid FORMAT: '*' may appear only before '('");
246         return 0;
247       }
248     }
249     ch = Capitalize(ch);
250     if (ch == '(') {
251       if (height_ >= maxHeight_) {
252         context.SignalError(IostatErrorInFormat,
253             "FORMAT stack overflow: too many nested parentheses");
254         return 0;
255       }
256       stack_[height_].start = offset_ - 1; // the '('
257       if (unlimited || height_ == 0) {
258         stack_[height_].remaining = Iteration::unlimited;
259         unlimitedLoopCheck = offset_ - 1;
260       } else if (repeat) {
261         if (*repeat <= 0) {
262           *repeat = 1; // error recovery
263         }
264         stack_[height_].remaining = *repeat - 1;
265       } else {
266         stack_[height_].remaining = 0;
267       }
268       ++height_;
269     } else if (height_ == 0) {
270       context.SignalError(IostatErrorInFormat, "FORMAT lacks initial '('");
271       return 0;
272     } else if (ch == ')') {
273       if (height_ == 1) {
274         if (stop) {
275           return 0; // end of FORMAT and no data items remain
276         }
277         context.AdvanceRecord(); // implied / before rightmost )
278       }
279       if (stack_[height_ - 1].remaining == Iteration::unlimited) {
280         offset_ = stack_[height_ - 1].start + 1;
281         if (offset_ == unlimitedLoopCheck) {
282           context.SignalError(IostatErrorInFormat,
283               "Unlimited repetition in FORMAT lacks data edit descriptors");
284         }
285       } else if (stack_[height_ - 1].remaining-- > 0) {
286         offset_ = stack_[height_ - 1].start + 1;
287       } else {
288         --height_;
289       }
290     } else if (ch == '\'' || ch == '"') {
291       // Quoted 'character literal'
292       CharType quote{ch};
293       auto start{offset_};
294       while (offset_ < formatLength_ && format_[offset_] != quote) {
295         ++offset_;
296       }
297       if (offset_ >= formatLength_) {
298         context.SignalError(IostatErrorInFormat,
299             "FORMAT missing closing quote on character literal");
300         return 0;
301       }
302       ++offset_;
303       std::size_t chars{
304           static_cast<std::size_t>(&format_[offset_] - &format_[start])};
305       if (PeekNext() == quote) {
306         // subtle: handle doubled quote character in a literal by including
307         // the first in the output, then treating the second as the start
308         // of another character literal.
309       } else {
310         --chars;
311       }
312       context.Emit(format_ + start, chars);
313     } else if (ch == 'H') {
314       // 9HHOLLERITH
315       if (!repeat || *repeat < 1 || offset_ + *repeat > formatLength_) {
316         context.SignalError(
317             IostatErrorInFormat, "Invalid width on Hollerith in FORMAT");
318         return 0;
319       }
320       context.Emit(format_ + offset_, static_cast<std::size_t>(*repeat));
321       offset_ += *repeat;
322     } else if (ch >= 'A' && ch <= 'Z') {
323       int start{offset_ - 1};
324       CharType next{Capitalize(PeekNext())};
325       if (next >= 'A' && next <= 'Z') {
326         ++offset_;
327       } else {
328         next = '\0';
329       }
330       if (ch == 'E' ||
331           (!next &&
332               (ch == 'A' || ch == 'I' || ch == 'B' || ch == 'O' || ch == 'Z' ||
333                   ch == 'F' || ch == 'D' || ch == 'G' || ch == 'L'))) {
334         // Data edit descriptor found
335         offset_ = start;
336         return repeat && *repeat > 0 ? *repeat : 1;
337       } else {
338         // Control edit descriptor
339         if (ch == 'T') { // Tn, TLn, TRn
340           repeat = GetIntField(context);
341         }
342         HandleControl(context, static_cast<char>(ch), static_cast<char>(next),
343             repeat ? *repeat : 1);
344       }
345     } else if (ch == '/') {
346       context.AdvanceRecord(repeat && *repeat > 0 ? *repeat : 1);
347     } else {
348       context.SignalError(IostatErrorInFormat,
349           "Invalid character '%c' in FORMAT", static_cast<char>(ch));
350       return 0;
351     }
352   }
353 }
354 
355 template <typename CONTEXT>
356 DataEdit FormatControl<CONTEXT>::GetNextDataEdit(
357     Context &context, int maxRepeat) {
358 
359   // TODO: DT editing
360 
361   // Return the next data edit descriptor
362   int repeat{CueUpNextDataEdit(context)};
363   auto start{offset_};
364   DataEdit edit;
365   edit.descriptor = static_cast<char>(Capitalize(GetNextChar(context)));
366   if (edit.descriptor == 'E') {
367     edit.variation = static_cast<char>(Capitalize(PeekNext()));
368     if (edit.variation >= 'A' && edit.variation <= 'Z') {
369       ++offset_;
370     }
371   }
372 
373   if (edit.descriptor == 'A') { // width is optional for A[w]
374     auto ch{PeekNext()};
375     if (ch >= '0' && ch <= '9') {
376       edit.width = GetIntField(context);
377     }
378   } else {
379     edit.width = GetIntField(context);
380   }
381   edit.modes = context.mutableModes();
382   if (PeekNext() == '.') {
383     ++offset_;
384     edit.digits = GetIntField(context);
385     CharType ch{PeekNext()};
386     if (ch == 'e' || ch == 'E' || ch == 'd' || ch == 'D') {
387       ++offset_;
388       edit.expoDigits = GetIntField(context);
389     }
390   }
391 
392   // Handle repeated nonparenthesized edit descriptors
393   if (repeat > 1) {
394     stack_[height_].start = start; // after repeat count
395     stack_[height_].remaining = repeat; // full count
396     ++height_;
397   }
398   edit.repeat = 1;
399   if (height_ > 1) {
400     int start{stack_[height_ - 1].start};
401     if (format_[start] != '(') {
402       if (stack_[height_ - 1].remaining > maxRepeat) {
403         edit.repeat = maxRepeat;
404         stack_[height_ - 1].remaining -= maxRepeat;
405         offset_ = start; // repeat same edit descriptor next time
406       } else {
407         edit.repeat = stack_[height_ - 1].remaining;
408         --height_;
409       }
410     }
411   }
412   return edit;
413 }
414 
415 template <typename CONTEXT>
416 void FormatControl<CONTEXT>::Finish(Context &context) {
417   CueUpNextDataEdit(context, true /* stop at colon or end of FORMAT */);
418 }
419 } // namespace Fortran::runtime::io
420 #endif // FORTRAN_RUNTIME_FORMAT_IMPLEMENTATION_H_
421